Masters Thesis

THE AMERICAN DREAM: HOME OWNERSHIP AND QUALITY OF LIFE INDICATORS IN NEW ORLEANS, LOUISIANA A Thesis Submitted to the Graduate Faculty of the University of New Orleans In partial fulfillment of the requirements for the degree of Master of Urban and Regional Planning in The College of Urban and Public Affairs by Zachary Klaas B. A., History and Political Science, Iowa State University, 1990 B. A., Philosophy, University of New Orleans, 1992 December 1997

i

For what are we but tenants for a day? -- Henry George

ii

ACKNOWLEDGMENTS

I would like to thank the following people for all their support and patience during the months I spent laboring on this thesis project; Janine Jean, my friend and best counsel throughout the period of my worst discouragement; Shannan Cvitanovic, who provided me with both friendship and shelter during the last month of this project; my friends and colleagues Bart Pittari and Kelvin Jackson, for helping me to keep the faith when most could see I was losing it; Yvan Lambert, for his assistance with the map and table layouts; the members of my thesis committee, Dr. Mickey Lauria, Dr. Raymond Burby and Dr. John Wildgen, without whose help this thesis would not have been possible given the necessary broadening of both my geographic and statistical skills; Dr. Vern Baxter, whose counsel at a crucial time during this thesis project gave me much-needed insight; Tim Joder and Dr. Fritz Wagner, whose assistance with none-too-minor bureaucratic problems made the completion of this thesis possible; Mike Reynolds and the New Orleans Police Department for the use of their city crime data; Dr. Denise Strong and Professor Jane Brooks, for their help in the formation of the thesis proposal; Katherine Hart, for seeming to understand what I was going through; and Denise Boswell, Denver Mullican, Steve Romano and Jerry Binninger, who were valued friends to me during the past year.

iii

TABLE OF CONTENTS

Acknowledgments Table of Contents List of Figures List of Tables Abstract

iii iv vi vii viii

Chapters I II III IV HOME OWNERSHIP AND QUALITY OF LIFE IN NEW ORLEANS HOME OWNERSHIP AND THE AMERICAN DREAM QUALITY OF LIFE RESEARCH HYPOTHESES Hypothesis One: Correlations Hypothesis Two: Factors Hypothesis Three: Scales 2 13 33 39 39 40 42

iv

V

DATABASE METHODOLOGY Constructing the Relational Database Indicator Construction Considerations Indicator Justification The Indicators

45 45 54 60 64 86 86 88 93 97 99 101 110 124 132 132 133

VI

STATISTICAL METHODOLOGY Correlations Factor Analysis Linear Regression Curvilinear Regression

VII

RESULTS Home Ownership and the Indicators Factoring of the Indicators Regressions on the Indicator Data

VII

CONCLUSIONS Theory Implications Policy Implications

Bibliography Vita

137 147

v

List of Figures Figure 1. Figure 2. New Orleans, Louisiana Percentage Home Ownership in New Orleans by Census Block Group Figure 3. Figure 4. Figure 5. Block Group Centroid Proximity to Libraries Geometric Concepts in Principal Components Analysis Median Household Income in New Orleans by Census Block Group Figure 6. Figure 7. Block Group Centroid Proximity to Hospitals Percentage Riding RTA in New Orleans by Census Block Group Figure 8. Figure 9. Figure 10. The Isolation of Lakeview Commuting Waves in New Orleans by Census Block Group Percentage of Vacant Units in New Orleans by Census Block Group Figure 11. Figure 12. Cubic Regression of Home Ownership by 36-Indicator Index Cubic Regression of Median Household Income by 36-Indicator Index 129 121 126 109 114 116 103 106 3 56 92 1

vi

List of Tables Table 1. Table 2. Table 3. Table 4. Table 5. Table 6. Table 7. Table 8. Table 9. Table 10. Table 11. Table 12. The Study Variable: Home Ownership Quality of Life Indicator Measures Levels of Spatial Aggregation Number of Cases in Geographic Datasets Indicator Variables Identified By Variable Construction Type Sources for Indicator Variables Indicators Correlation Matrix with Percentage Home Ownership Indicators Correlation Matrix with Median Household Income Block Group Level Correlation Matrix from Factor Analysis Tract Level Correlation Matrix from Factor Analysis Neighborhood Level Correlation Matrix from Factor Analysis Index Correlations for Home Ownership and Median Household Income Table 13. Factor Correlations for Home Ownership and Median Household Income 130 125 8 8 10 47 48 53 102 105 112 119 123

vii

ABSTRACT

The existence of a strong causal relationship between high levels of home ownership and a high quality of life in urban neighborhoods is presumed by public policy in many American cities. In New Orleans, Louisiana, this presumption that home ownership increases the quality of life in neighborhoods is the prime basis of the city's housing policies. This presumed relationship is also a principal justification for the efforts of a number of non-profit housing organizations, which seek to promote home ownership as a form of tenure, particularly in lower- and moderate-income neighborhoods. These non-profit organizations (which include the New Orleans Home Mortgage Authority, the Neighborhood Development Foundation and Neighborhood Housing Services) have joined the city in directing their efforts at promoting neighborhood quality of life through establishing home ownership as the prevailing form of tenure, despite the fact that there has been no attempt to validate whether home ownership has the presumed effect of promoting neighborhood quality of life. In this thesis, I explore the benefit increasing levels of home ownership has in terms of indicators of quality of life at three geographic scales, in order to subject the policy presumption concerning home ownership to a test of falsifiability. In particular, I elaborate the difference between the correlative relationship of an area's home ownership on quality of life indicators and the correlative relationship of an area's median household income on quality of life indicators. Since one's propensity to own a home (in the

viii

absence of the efforts of the city and non-profit housing organizations) is likely to be a function of one's income level, any difference between home ownership's benefit in terms of the quality of life indicators and median household income's benefit in terms of those same indicators is of great theoretical importance. The construction of indicators of quality of life poses in itself a fairly new problem in the urban planning and geography literatures. Though geographic information systems technology has made the analysis of neighborhood-level data feasible and cost-effective, the difficulty of determining what properly counts as a true measure of quality of life has been an obstacle in the creation of indicator studies. However, the similarity of indicators used in many studies which have appeared in recent years suggests that "quality of life" is not necessarily a wholly subjective notion. I make an effort to use the more ubiquitous and prevalent indicators from well-known studies in the consideration of the research question. This analysis does not make use of psychological “sense of well-being” measures, due to the extensive survey research this would require. The analysis is, in this sense, somewhat incomplete, and future research into this topic should incorporate such measures to gain further insight into the more individual and perceptive aspects of quality of life. In order to further clarify what sorts of indicators are affected by home ownership, the indicators are grouped into factors via factor analysis and each factor is also regressed against home ownership and median household income. A theoretical expectation is that

ix

there will be differences in how home ownership correlates with certain factors at certain geographic scales. The data for this project were analyzed using the ArcView 3.0a geographic information systems software made by the Environmental Systems Research Institute (ESRI), the Statistical Package for the Social Sciences (SPSS-PC) Version 7.0 statistical software, the Microsoft Excel 97 spreadsheet program and the Microsoft Access 97 relational database program.

x

1

Figure 1. New Orleans, Louisiana

2

CHAPTER I HOME OWNERSHIP AND THE QUALITY OF LIFE IN NEW ORLEANS

Americans have long equated home ownership with a better quality of life, indeed, more so than perhaps any other nation. The common use of the phrase "the American Dream" as a synonym for the cultural institution of home ownership alerts one to this fact possibly more than anything. The opportunity to own a home has been venerated by generations of Americans as a hallmark of freedom and independence; the capacity to do so has been equally venerated as a sign of social respectability, economic strength and commitment. Consequently, city governments and non-profit housing organizations throughout the United States have long sought to increase the home ownership rates in the nation's urban areas. The city of New Orleans, Louisiana (Figure 1), like so many other cities in the United States, has exhibited a strong policy preference for increasing its number of home owners. According to the 1990 U. S. Census, only 36.5% of the total housing stock of the city of New Orleans was owner-occupied housing. Representatives of the city administration find this figure low, despite the fact that New Orleans has historically never had a large percentage of home owners; in their estimation, the high proportion of rentership accounts for a great deal of the developmental malaise in which the city finds

3

Figure 2. Percentage Home Ownership in New Orleans by Census Block Group (quintile distribution)

4

itself, and increasing ownership would have a clear benefit in terms of the quality of life in the city's neighborhoods. The distribution of home ownership in the city (Figure 2) shows that rentership is concentrated in the poorer and less well-maintained central part of the city, while owner-occupancy is concentrated in the affluent and well-maintained Lakefront, Uptown and Lower Algiers areas of the city. This policy judgment, that increasing home ownership would clearly and positively affect the quality of life in the New Orleans neighborhoods, has in recent years led to the city's involvement with the promotion of below-market interest mortgage loans for first-time home buyers. Working with three non-profit housing organizations, the city has acted to promote home ownership through the offering of such loans to lower- and moderate-income individuals and families who might otherwise not be able to afford to purchase a home. The three non-profit organizations are: the New Orleans Home Mortgage Authority (NOHMA), directed by Mr. Mtumishi St. Julien; the Neighborhood Development Foundation (NDF), directed by Ms. Rosalind Peychaud; and Neighborhood Housing Services (NHS), directed by Ms. Lauren Anderson. These three non-profit organizations facilitate mortgage transactions with lower- and moderate-income borrowers. In addition, they provide all applicants with "home ownership training," a formal education in the responsibilities of home ownership which is prerequisite to final approval for the reduced-interest loan.

5

Through subsidized mortgage rates, the city is able to offer a choice to those who have long desired to own a home but were priced out of the market; by doing so, of course, the city interferes with the normal functioning of the housing market, inflating demand and stimulating supply for owner-occupied housing. The benefits of this staying of Adam Smith's "invisible hand" are taken, in the view of the city and the non-profit housing organizations, to outweigh the costs. Home ownership is taken to be of such paramount importance to the quality of life in New Orleans neighborhoods, that it is worth the necessary government interference with the natural functioning of the economy. Such a policy judgment rests upon a largely untested presumption; it rests upon the presumption that home ownership actually affects the quality of life positively in the first place. This assumption, made quickly in a country which sees home ownership as the "American Dream," has been taken to be such an obvious one that the relevance of even testing the notion is not apparent to many. Indeed, home ownership is often considered evidence of a higher quality of life in and of itself, because it is presumed that Americans desire so greatly to own their own homes that their lives would be better simply if they did, regardless of any other considerations. Since other considerations do not need to be made, by such reasoning, they are usually not made, with the result being that the quality of life benefits of home ownership are taken for granted rather than explored and questioned.

6

My objective in this master's thesis is to explore the relationship of the percentage of home ownership to indicators of neighborhood quality of life in an attempt to determine whether home ownership is the unalloyed policy treasure its supporters would make it out. In order to accomplish this, there will be some effort expended in the following pages to design and implement measures of neighborhood quality of life for the city of New Orleans. Since the advent of geographic information systems (GIS) technology, there has been increased interest in the maintenance and use of neighborhood indicator databases. In the past ten years in particular, the number of city planning departments which have done GIS-based indicator studies has burgeoned, given the increasing simplicity which modern GIS software lends to the task. This increase in the analytic capacity of city planning departments has engendered a renewed interest in the theory behind the design and implementation of quality of life indicator measures. Neighborhood organizations and academics have also benefited from this increased analytic capacity, and the past decade has also seen an increase in the creation of indicator measures by citizen groups, urban geographers, sociologists and planners. Within the last five years, there has been considerable cross-fertilization of ideas between city planning departments and these neighborhood groups and academics, and the methodology employed in the design of indicator measures has come to be fairly similar. Though there is some variability in indicator measures from place to place, with some

7

indicator measures only making sense in certain contexts, there are a number of indicator measures which seem to be ubiquitous and commonly selected by city planners, neighborhood groups and academics alike. Some of the more commonly used indicators rely upon survey response data and present a psychological perspective of quality of life as it is experienced by the individual. An effort was not made to fully administer an opinion survey to residents of New Orleans neighborhoods for the purposes of this thesis, but such surveys are valuable for the construction of “sense of well-being” indicators. Further studies utilizing such indicators would extend our knowledge of quality of life in New Orleans, particularly considered as a more experiential, psychological concept. Based on recent studies, a number of commonly used measures were selected for use for the purposes of this thesis. One of these measures is our study variable, the percentage of home ownership (Table 1). The following indicators (Table 2) either appear frequently as part of a number of previous indicator studies or represent domains of inquiry which are common in previous indicator studies. Each of the indicator variables will be defined for the reader at length in the chapter of this thesis dealing with database methodology; the justification for the use of each variable will also be provided in that section. The research question to which this thesis addresses itself is whether home ownership affects neighborhood quality of life indicators, positively or negatively. If it

8

Table 1. The Study Variable: Home Ownership Name Source Percentage of Hom e Ownership U.S. Census Table 2. Quality of Life Indicator Measures Name Bicycling To W ork CAT Elem entary Test 50th Percentile+ Clothing Stores Com m uting Index Diversity of Em ploym ent Doctors Doctors Em ploym ent Fire Stations Graduate Exit Exam 90%+ Grocery Stores Grocery Stores Health Risk High School Graduation Rate Hom e Affordability Hospitals LEAP Elem entary Test 90%+ LEAP Jr./Middle Test 90%+ Libraries Median Household Incom e Median Property Value Nonviolent Crim e Parks Parks Police Stations Post-High School Graduation Rate Public Assistance Real Em ploym ent Recreation Facilities RTA To W ork Schools Toxic Releases Toxic W ater Releases Traffic Index Vacancies Violent Crim e Violent Non-Property Crim e W alking To W ork

Source U. S. Census New Orleans Tim es-Picayune Switchboard Phone Num ber Query W ebsite U. S. Census U. S. Census Switchboard Phone Num ber Query W ebsite Switchboard Phone Num ber Query W ebsite U. S. Census New Orleans Fire Departm ent New Orleans Tim es-Picayune Switchboard Phone Num ber Query W ebsite Switchboard Phone Num ber Query W ebsite Toxic Release Inventory U. S. Census U. S. Census Switchboard Phone Num ber Query W ebsite New Orleans Tim es-Picayune New Orleans Tim es-Picayune Switchboard Phone Num ber Query W ebsite U. S. Census U. S. Census New Orleans Police Departm ent New Orleans City Map New Orleans City Map New Orleans Police Departm ent/Metropolitan Crim e Com m ission U. S. Census U. S. Census U. S. Census Switchboard Phone Num ber Query W ebsite U. S. Census Switchboard Phone Num ber Query W ebsite Toxic Release Inventory Toxic Release Inventory U. S. Census U. S. Census New Orleans Police Departm ent New Orleans Police Departm ent U. S. Census

9

can be shown that increased levels of home ownership does not seem to be positively correlated with indicators of neighborhood quality of life, then the extensive efforts the city makes to promote home ownership will seem less warranted, or at least less warranted to promote the ends the city may have had in mind. Indeed, if home ownership correlates negatively with neighborhood quality of life indicators, then promotion of home ownership, in the absence of policies which control for any deleterious effects increased home ownership might itself cause, would run contrary to the articulated goals of city policy. Part of the research question, if I am to adequately address how increased levels of home ownership affects neighborhoods, involves identifying which indicator measures seem to behave in similar ways with other indicator measures. Through the statistical method of factor analysis, groups of indicators which behave in similar ways in different neighborhoods can be fashioned from indicator data, and these groups can be posited as underlying quality of life "factors." To explain this idea another way, if indicators which all come from a certain identifiable domain (e.g., "education indicators" or "indicators of distance to basic services") also tend to have similar values relative to each other in different neighborhoods, then this is evidence there is an underlying factor which these indicators are both measuring. Factor analysis allows for greater understanding of how

10

home ownership affects certain categories of quality of life indicators, as well as for a greater understanding of how much specific social, economic, environmental or cultural forces may be affected by increased levels of home ownership. Another consideration to which this research project addresses itself is that of geographic scale. The neighborhood quality of life indicators constructed for this thesis are aggregated at three different spatial levels of aggregation (Table 3).
Table 3. Levels of Spatial Aggregation Level Source Block Group Tract Neighborhood U. S. Census U. S. Census City of New Orleans, Data Analysis Unit

Each of the preceding are polygonal geographic shape coverages within the city of New Orleans which, for the purposes of this thesis, define an area for which the level of home ownership is known and in which the indicator values may be calculated. The first two coverages are designations used by the United States Census, while the third is a designation created by the city of New Orleans in 1983 and still used by the city as a formal operational definition for "neighborhood." There is no overlap between polygons in any of these spatial levels of aggregation; a census tract is composed of census block groups, and neighborhoods are composed of census tracts. Just as it is of some importance whether certain groups of indicators are affected by home ownership in different ways than others, it is of some importance whether certain indicators are affected by home ownership at certain geographic scales more than

11

others. If, for example, home ownership seems to correlate positively with quality of life indicators at the block group level but not at all at the tract level and negatively at the neighborhood level, then the conclusion would be that a city policy of increasing home ownership would be constructive at the block group level, but unimportant at the tract level and actually destructive at the neighborhood level. The result of such a policy would be the promotion of balkanized subdivisions rather than neighborhoods. One final comparison must be made part of the research design for the project to produce meaningful results, namely, a comparison of how other variables which are highly collinear with home ownership relate to the increases in quality of life indicators. In the absence of reduced-rate mortgage programs, home ownership is generally a function of having the necessary income to purchase a home. Thus, the intercorrelation between home ownership and median household income may properly be expected to be fairly high, an important task will be to see how income relates to the remaining indicator variables and compare it to how home ownership relates to those same variables. If home ownership correlates more highly with the indicators, then the defenders of current city policy will have a statistic for their arsenal. If income correlates more highly, then it would be fair to say that any positive benefit regarding quality of life stemming from home ownership can be better explained by the income level of those who own the homes.

12

The research question with which this thesis concerns itself may now be formulated as follows: "How much do areas with a high percentage home ownership benefit from a high quality of life, taking into consideration factors underlying quality of life and geographic scale?" The focus of inquiry having been thus defined, the next chapter of this thesis proceeds to a review of the home ownership and quality of life indicator literatures, which will provide a background for the proposed research hypotheses. I will deal with two topics, home ownership and quality of life indicators, each of which have their own separate literature. These two literatures are therefore reviewed in separate chapters. The chapter following will formally state the research hypotheses, and the ensuing two chapters will describe the methodologies used to construct the database for the project and to statistically analyze it. The final two chapters will set forth the results of the inquiry and draw conclusions from the analysis.

13

CHAPTER II HOME OWNERSHIP AND THE AMERICAN DREAM

Home ownership is often referred to as "the American Dream," and the ability to purchase a home is often seen by Americans as a great triumph of political and economic independence for the individual who is so fortunate as to possess it (Zundel, 1995). The ownership of one's home is also seen as evidence of commitment to one's community, as a spur to greater sociability and better citizenship (Tremblay and Dillman, 1983:37), as a demonstration of concern for the physical maintenance of the immediate environs (Galster, 1987; Tremblay and Dillman, 1983:39), and as the prerequisite to true social equality (Heskin, 1984). Such beliefs resonate strongly not only in American culture, but in the cultures of all nations with predominant Anglo-Saxon ethnic backgrounds (Kemeny, 1981). The reason for this lies with home ownership's connection with cultural values regarding land ownership which were prevalent in Great Britain for most of the 17th and early 18th centuries and which spread to her colonies during this period. The political economic reality of Britain during this time was such that those who owned land also controlled politics, since the economy was primarily agricultural. Feudal lords and the landed gentry were the primary political movers during this time, and the contemporary English were quick to recognize the connection between land ownership and political power. Since those who did not own land were economically

14

dependent on those who did, reasoned James Harrington, a well-known 17th century English political philosopher, political obedience to those who controlled the means of economic production was necessary. The ownership of land by individual freeholders or yeoman farmers, for Harrington, was the key to political independence (Harrington, 1992 [1656]; Harrington, 1992 [1661.]) Harrington's ideas were similar to those propounded by a British political party called the Levellers during the time of the short-lived English republic; John Lilburne advocated a more equitable distribution of land as leader of the Levellers in the House of Commons before being executed by Oliver Cromwell (Frank, 1955). Though Lilburne was defeated in his efforts, his notions of "land and liberty" continued to strike a chord that was to resound not much later in England's history. The political philosopher John Locke picked up the notion in his Second Treatise On Government (Locke, 1967 [1699]), in which he defended the right of freeholders to claim unclaimed land on the condition that they cultivate the land and leave "as much and as good" for others to use. Locke viewed the right to claim absolute exclusive ownership of the land as necessary for liberty, and his elaborate defense of property rights can properly be seen as a strong defense in particular of the right to own land. The ideas of James Harrington and John Locke were extremely influential among the American colonists of the 18th century, for obvious reasons; in the New World, unclaimed land was plentiful, and for those willing to bear the burden of cultivating the

15

land and building a house upon it, the claiming of this land meant freedom from the domination of Europe's feudal nobility and landed gentry. Locke's ideas in particular were a strong influence upon Thomas Jefferson, who envisioned America as an agrarian republic where individuals were empowered by their self-sufficiency as agricultural producers to lead lives free of subjugation to others (Appleby, 1984; Jefferson, 1984 [1785]; Kim, 1978; Zundel, 1995). Jeffersonians throughout the new nation of the United States began to see industrialization as a threat to the "agrarian republican" vision of freeholding yeoman farmers. The contemporary American political philosopher John Taylor of Caroline inveighed in his writings against the servitude implied by wage-earning and the paying of rent. True freedom belonged, he argued, to the farmer, whose ownership of the means of production, the means of his sustenance and shelter placed him on an equal footing with anyone. The urban worker, on the other hand, neither owned the means of production nor sustenance, and even had to pay rent to inhabit a domicile. Thus he was wholly dependent upon the good will of his employer and his landlord, and subservient to both (Sellers, 1991; Taylor, 1818). Taylor's view of agricultural land ownership as the safeguard of a free republic is well exposited in the following passage from Arthur Schlesinger's The Age of Jackson: Like Jefferson, Taylor was possessed by a vision of a free society; and like Jefferson, he was profoundly aware of the material conditions of freedom. In his own emphatic words, "Wealth, like suffrage, must be considerably distributed, to sustain a democratick republic; and hence,

16

whatever draws a considerable proportion of either into a few hands, will destroy it. As power follows wealth, the majority must either have wealth or lose power." In the continuing supremacy of agriculture - "the guardian of liberty as well as the mother of wealth" - he saw the best promise for a wide distribution of property. Agriculture dealt in what he called "natural" or "substantial" property, that is, property directly gained in productive labor. It was his central thesis that the preservation of a free democracy depended on the preservation of the order of "natural" private property, where each man was secure in the fruits of his own labor, against depredations by force or by fraud (Schlesinger, 1945: 22-23). By this sort of reasoning, an outright attack on landlords, who withheld from individuals the full proceeds from their labor by charging rent, seemed clearly called for, and there was a consequent justification for attaching a stigma to tenancy and rentership (Dreier, 1982; Heskin, 1984). The "Anti-Renter" movement of the early 18th century harkened to the Jeffersonian agrarian ideal in its both its advocacy of the right of ownership and its demand for the opportunity of ownership; this movement was the precursor of a larger movement, the "farmer-labor" movement in American politics, which united farmers weaned on the ideas of John Taylor of Caroline and conscious of their importance in an agrarian republican America with urban workers increasingly aware of the advantage taken of them by employers and landlords (Rice, 1969). A main rationale for the yeoman freeholder existence was freedom from the market (Sellers, 1991: Chapter 1). The life of a freeholder, as idealized by Jeffersonians, was characterized by self-sufficiency and communal sharing of resources. Independence

17

from sullying market forces and ties to creditors was the goal of Jeffersonian economic policies. During the presidency of Andrew Jackson, the freeholder way of life was protected by the arch-Jeffersonian Jackson against banking interests and the pro-market Whig party (Schlesinger, 1945; Sellers, 1991). As market capitalism came to replace agrarianism throughout the country in the later part of the 19th century, the practical importance of land ownership began to change. In the new order of things, the ownership of increasingly scarce prime agricultural soil was seen to be of less importance than the ownership of physical structures on the land - in short, land ownership gave way to home ownership as the primary necessity for those desirous of liberty. Much of the previous justifications for antagonism to landlords were undercut with the lessened emphasis on agricultural production. John Taylor of Caroline, for example, railed against the notion of paying rent because owning the land allowed a farmer control over the means of production. Home ownership, by contrast, would do nothing to help anyone gain control over the means of production. Likewise, ownership of land helped to provide sustenance for the owner. Owning a home, by contrast, would do nothing to help provide the owner's daily bread. Contemporary writers have noted that the agrarian republican sociological "frame" rather inadequately addresses this historical change in its assessment of the importance of ownership (Zundel, 1995).

18

Nevertheless, there were aspects of home ownership which rendered the transference of these previous attitudes concerning land ownership reasonable. The political economists of the 19th century had their own insights into rentership, and as their works gained currency, Americans found new reasons to hate their landlords and long for a home of their own. English political economist David Ricardo made the following observation: On the first settling of a country, in which there is an abundance of rich and fertile land, a very small proportion of which is required to be cultivated for the support of the actual population, or indeed can be cultivated with the capital which the population can command, there will be no rent; for no one would pay for the use of land, when there was an abundant quantity not yet appropriated, and, therefore, at the disposal of whosoever might choose to cultivate it (Ricardo, 1911 [1817]: 34). Such an observation might equally well apply to home ownership in a new country - so long as there is enough timber in unclaimed forests, no one would pay for a house one could as easily build oneself from abundant raw materials on a plot of unclaimed land. Ricardo noted, however, that rent could be charged as soon as there was scarcity, and those lucky enough to own the prime land would reap the benefits of this scarcity; they would profit simply by owning the land. This, too, was equally the case for home ownership, and in America this became clearer after the closing of the American frontier in the last decades of the 19th century (Turner, 1920). Ricardo's view was that the economic interests of landlords, given the fact that rent increases proportionally to the scarcity of good land, were necessarily opposed to that of the remainder of society. The

19

higher the rents taken in by the landlord, the worse-off and more dependent everyone else would be. The American political economist Henry George took this observation to its logical conclusion in his major work Progress and Poverty. George took Ricardo's observations and adapted them for a post-agrarian setting. George's main contribution to political economy, as many know, is his advocacy of a "single tax" on landlords, whom he blamed for the economic dysfunctions of 19th century America. Revealing, however, are George's theoretical reasons for supporting such a tax. Like Ricardo, George believed that the economic interests of landlords were necessarily inimical to the remainder of society: If one man owned all the land accessible to any community, he could, of course, demand any price or condition for its use that he saw fit; and, as long as his ownership was acknowledged, the other members of the community would have but death or emigration as the alternative to submission to his terms (George, 1966 [1879]: 68). However, George's definition of "land" differed considerably from Ricardo's. Both Ricardo and George agreed that "land" was one of the three sectors of political economy, the other two sectors of which were "labor" and "capital." Unlike Ricardo, however, George defined "land" as any resource which exists in the same form in nature. Thus, for George, "land" could be the soil a tree grows in, but it could also be the tree, or the wood from a tree. In Georgist terminology, "land" meant any purely natural resource: The term land embraces, in short, all natural materials, forces and opportunities, and, therefore, nothing that is freely supplied by

20

nature can be properly classed as capital (George, 1966 [1879]: 12). Thus, for George, rent on "land" referred to a payment made to an individual for something that nature, and not that individual, had done. George felt this to be unconscionable and therefore opposed rent as an "unearned increment" reaped on nature's bounty. Notably, he was careful to distinguish what he called "rent" from any money which reimbursed a property owner for improvements made to the property or for payment of labor for construction. For George, "rent" meant only that payment demanded simply for the use of resources which belonged to someone else. The right to exclusive ownership of anything of human production is clear...[t]o improvements, such an original title can be shown; but it is a title only to the improvements, and not to the land itself. If I clear a forest, drain a swamp, or fill a morass, all I can justly claim is the value given by these exertions (George, 1966 [1879]:158). This distinguishing of an evil "rent" from a wholly proper reimbursement for managerial services rendered in the production or maintenance of housing is what made George's proposed solution of the "single tax" so cumbersome and unrealistic; it is often very difficult to determine what constitutes a "fair return" on a landlord's investment in a rental property. Rent control measures recently applied in a number of U. S. cities have included estimations of fair return for the landlord's investment (Gilderbloom, 1981; Goetz, 1995). These measures, called "second generation" or "moderate" rent controls because of their inclusion of such estimations, are evidently a strong improvement over the previous generation of rent control measures, which are routinely featured in college

21

economics texts as examples of how not to regulate the housing market. Nevertheless, there has been no settled-upon way of using estimations to separate a fair return on a landlord's investment from George's "unearned increment." One way around the problem, of course, would be to buy one's home rather than renting it from a landlord in the first place, thus taking the potential exploiter out of the picture. Americans, already suspicious of landlords from the days of agrarian republicanism, became even more suspicious of them in the waning days of the Gilded Age, especially as the number of unhealthy and crime-ridden slum tenements began to burgeon in the nation's major cities. But despite these suspicions, most Americans could not purchase their own home because, in the days before the U. S. federal government offered banks insurance so they might be able to offer 30-year mortgages with a minimum of risk, most Americans simply did not have the money. The final development in America's cultural commitment to home ownership, then, was the outright promotion of home ownership by the federal government. This policy developed from Depression Era necessity as much as from free choice; the Federal Home Loan Bank Act of 1932 and the Housing Act of 1934 were intended as responses to the collapse of the nation's private mortgage financing system (Hays, 1985; Stone, 1971). Federal mortgage insurance guarantors such as the Federal National Mortgage Association (FNMA or "Fannie Mae," established 1938), the Federal Housing Administration (FHA, established 1944), the Veterans' Administration (VA, established

22

1944) were established to protect lending institutions against extreme losses on outstanding mortgages. With this protection, banks would not have to freeze their assets after a spate of failed mortgage loans, and thus these assets would be available for future mortgage loans; thus, a momentary downturn in the housing market would not begin a spiral into a complete collapse. In later years, other guarantors such as the Government National Mortgage Association (GNMA or "Ginnie Mae," established 1968) and the Federal Home Loan Mortgage Company (FHLMC or "Freddie Mac," established 1970) would be created in addition to the original agencies. These agencies, like a number of the "alphabet soup" agencies of the Roosevelt Administration, were intended as temporary, Keynesian "pump priming" measures which would correct the market and restore its essential self-regulation. A number of factors conspired to keep federal mortgage insurance guarantees politically afloat well after the demise of many of the temporary Depression Era measures. Due to the end of the Second World War and the return of millions of soldiers from Europe and the Pacific, there were now roughly 6 million men and women who needed to be housed. In addition, the housing industry, which had feared government intervention into the housing market, was now satisfied that the government subsidization of home ownership worked in their favor, and did its best to maintain the new status quo. Another brick in the wall of this new status quo in housing policy was the FHA federal income tax credit on mortgage interest and property taxes. This new credit

23

allowed individuals considerable tax breaks simply for being home owners (Hays, 1985:85). Since it applied to property taxes as well as mortgage interest, the credit in effect distributed monies collected through the federal revenue to local property tax jurisdictions. This gave local government a strong interest in home ownership, as it allowed municipalities to collect a good portion of their own revenue through no cost to the local taxpayer. Conservatives in Congress had originally expressed a fear that long-term, lowinterest mortgage guarantees and interest subsidization would "effect a regimentation of home-finance and home ownership patterns that would lead to a welfare state" (Wright, 1981: Chapter 6). This opinion, somewhat widely-held at the time of the passage of legislation establishing FNMA, FHA and VA, virtually disappeared from mainstream conservative political thought a mere twenty years later. One reason for this was the spur increased mortgage lending had provided for the process of suburbanization. The Housing Act of 1949, in response to the post-war housing crisis, set a goal of "a decent home and a suitable living environment for every American family" and guaranteed builders and bankers more substantial profits on large residential developments. In the wake of this legislation, suburbs bloomed all over the United States, and the new residents of those suburbs began to vote for conservative politicians (Donaldson, 1969; Wood, 1958). There is considerable debate about the antecedents for the rise in suburban conservative voting behavior in this period. Some

24

theories hold that the bland design and careful segregation of many suburbs by restrictive covenants and "converted" new suburbanites to conservative views and voting behaviors. Others hold that city-bound conservatives voluntarily "transplanted" themselves to the suburbs as more congenial locations for those with their outlook. Whatever the reasons for the change, conservatism itself changed greatly with the rise of the suburbs, and one major facet of the change was that conservative Republicans were now to act to promote the suburban lifestyle, even at the cost of supporting government intervention to do so. Another reason the new status quo in housing policy contributed to the new conservatism was that it rewarded “market winners” (Tremblay and Dillman, 1983: 37) such as developers and home builders, rather than poorer strata of society which could be assumed to have a more liberal economic philosophy. The large loss of revenue at the federal level due to mortgage insurance protection and the FHA tax credit “was not questioned by either liberals or conservatives” (Hays, 1985: 85), and a new normative consensus was fashioned in the process. Charging taxes differentially on the basis of housing tenure, which simply put is what the FHA tax credit does, can be properly seen as an act of normative conditioning. Sociologists Kenneth Tremblay and Don Dillman see the credit as a sanction against renters: As is the case with any norm, there exist sanctions to encourage conformity to the norm of home ownership. One sanction consists of the federal income tax laws that make interest on mortgage

25

payments and property tax payments tax deductible (Tremblay and Dillman, 1983: 36). Programs which a fiscal conservative might have criticized in the days before the New Deal as an unwarranted interference with the natural processes of the market, a cultural conservative could, by this interpretation, defend as promoting the social cohesion of a virtuous home owning middle class, or a guard against the troublesome vices of the tenement-dwelling poor. The overall change in mainstream conservative opinion continues to manifest itself today; even the most strident defenders of laisser faire, such as former HUD Secretary Jack Kemp, can find their way clear to continued and aggressive support of government assistance for first-time home buyer, even to the point of advocating that millions of dollars in federal block grants be targeted towards the conversion of public housing units to homes owned by their residents (Kemp, 1990; Rohe and Stegman, 1992). Indeed, the conservatism of former Secretary Kemp seems of a kind with the old agrarian republicanism of John Taylor of Caroline if one goes by such quotes as the following: [D]emocracy can't work without the component that goes to the heart of what freedom is all about - the chance to own a piece of property. (Raspberry, 1990). During Secretary Kemp's term as the director of the Department of Housing and Urban Development, Project HOPE (Home Ownership for People Everywhere) was inaugurated under the provisions of the National Affordable Housing Act of 1990. Project HOPE was designed to help low-income families buy public or otherwise

26

federally-owned, financed or insured units, offered to selected tenants at below-market rates. The amount of implicit federal subsidy involved in the offering of the properties at below-market rates was a principal reason for the program's apparent success at building the accumulated wealth of those who participated in it, according to William Rohe and Michael Stegman: The findings [of this evaluation study] provide limited support for...arguments in favor of public housing home ownership programs...[T]he program will assist many of the participants to accumulate wealth. Many will have considerable amount of convertible equity in their homes after a relatively short period of time. Yet most of this equity is the result of the below-market sales price, not of appreciation and the pay down of mortgage principal payments as is the case with typical home buyers. In essence much of the increase in wealth is a federal gift to the participants (Rohe and Stegman, 1992: 153). Not surprisingly, being the recipients of such federal largesse, those participating in the project tended to rate it rather highly, with 78 percent of respondents to a survey questionnaire stating that "owning a home had improved their self-concept" and roughly two-thirds reporting that home ownership had made them "more financially secure." After the experience of agrarian republicanism, the vilification of landlords by activists and intellectuals, and the nation's policy decision to guarantee home mortgage loans and subsidize the home building industry, America came to a point where it regarded home ownership not only an option but indeed a requirement. Home ownership is, indeed, considered by many Americans a rite of passage to adulthood (Perrin, 1978: Chapter 2). Various scholars have similarly identified the central importance of home

27

ownership in American culture, describing it as a public policy "ideology" (Hays, 1985) or a "myth" (Kemeny, 1981). It is perhaps one of the better examples of where strong sociological norms prescribe a conformity of economic behavior (Jones, 1984; Tremblay and Dillman, 1983: Chapter 2). It may perhaps not even be going too far to compare a "home owner ethic" to the famed "Protestant work ethic," which also turned economic accumulation into a normative duty (Weber, 1985 [1905.]) The very description of home ownership as "the American Dream" suggests that there is something peculiarly un-American about not owning a home. The National Association of Home Builders plays on this theme in its advertisements: Home ownership is the cornerstone of family security, stability and prosperity. It strengthens the nation's communities, encourages civic responsibility and provides a solid foundation from which Americans can work to support their families, enhance their communities and achieve their personal goals (National Association of Home Builders, 1997). The preceding passage is anything but timidly phrased; home ownership is not merely one of many possible means to "family security, stability and prosperity," but a "cornerstone" for these. Home ownership strengthens community, encourages civic responsibility and supports families; by force of argument, one can only assume that failure to own a home fractures community, discourages civic responsibility and weakens families. It is not difficult to carry from such rhetoric the implication that one who chooses to rent cares little about the rest of society.

28

Is this, however, as obvious a conclusion as the passage makes it seem? Advocates of home ownership have frequently made an extended case in the literature that it is. The main assertions include: · Home ownership allows individuals to save money by building equity in the home. Renters, when they pay their landlords, lose the money they pay; owners, by contrast, can accumulate equity (the difference between the value of the house and the amount left on the mortgage note) in the property. Thus home ownership is like a business investment which pays dividends with increases in property values. Much of mathematical housing market analysis proceeds from assumptions about equity-maximization, analyzing housing choice through "rational actor" econometric models (Clark and Moore, 1980). The presumed rise in housing values, of course, is of critical importance here, as equity could indeed become negative should property values in an area drop. One misconception of ownership, according to some scholars, is that a person can truly be said to own the home one is said to have equity in until one fully pays the bank. Should property values fall, equity is revealed to be the abstraction it essentially is (Stone, 1971). · Home ownership is a sign of commitment to a community. Renters may leave a distressed neighborhood at any time they choose without fear of a loss in

29

property investments, and thus may be seen as "transients" with only a passing concern for the upkeep and maintenance of the community. Owners are considered to have a stronger concern for community maintenance on the basis of their financial interest in the community (Galster, 1987). The flip side of this notion is, naturally, that renters' economic fortunes are not entirely tied down to the health of a particular urban neighborhood, and rentership allows them a greater degree of mobility in this kind of circumstance. In the words of Doris Lessing, "[w]hy should we spend all the capital we are ever likely to have tying ourselves down to a place we detest?" (Edel, Sclar and Luria, 1984: 188). Friedrich Engels argued in his On The Housing Question that home ownership not only chained urban workers "once again to the soil," but further more compelled them to "put up with whatever working conditions are offered to them" because they have house payments left to make and cannot therefore cannot risk unemployment. Thus could home ownership, in fact, actually serve as a repressive method of social control (Engels, 1975 [1872]). Most evidence in the United States indicates that home ownership is not seen by most Americans as a method of repression, however much discontent may be associated with day-to-day difficulties involved with owning a home (Edel, Sclar and Luria, 1984:171-89). · Home ownership promotes pride and a feeling of ownership. That there is some psychological value in home ownership for members of a culture that

30

values home ownership is both an undeniable and clearly circular proposition. However, outside of America, there does appear to be some level of satisfaction with rental or co-operative tenure patterns (Kemeny, 1981:7-10). An added facet of this psychological defense of home ownership is its claim that people prefer living in detached structures with large lots. This seems at least somewhat rooted in the American experience as well, given the large social housing sectors in many European countries and the attached structures which predominate in those countries. Nevertheless, most Americans clearly do prefer detached housing with wide lots (Tremblay and Dillman, 1983:3739). · Home ownership gives an individual a stake in society and thus encourages civic responsibility and sociability. The essence of this claim is that a home owner, given a financial stake in the level of property values in an area, will have an incentive to work with others in the community in order to safeguard it from threats to those values. Thus, the home owner will be a joiner of organizations and will learn the necessary social skills to work with neighbors and fellow property owners to protect common interests. Renters, by contrast, could remain socially aloof and would have no incentive to join community

31

groups or work toward a shared goal with other members of the community (Dreier 1982). A contrasting picture of home owner behavior can, of course, be drawn by referencing the racism which has commonly been associated with community groups that join together to protect neighborhood property values. FHA policies originally overtly encouraged ethnic and racial exclusion through restrictive covenants in the name of maintaining property values (Wright, 1981: Chapter 6). Civic responsibility behavior among renters is also somewhat inhibited by widespread prejudice against renters in neighborhood-level civic groups and associations, justified by the perception on the part of owners that renters are insufficiently committed to their neighborhoods (Dreier, 1982). · Home ownership deters crime, misbehavior and deviance. This claim does not, on the surface, seem so much to be about form of tenure as about the design principles which guide the creation of neighborhoods that are mostly single-family detached homes. The comparison is made to multi-family attached rental areas which encourage people to congregate in central locations and thus provide criminals with opportunities. Single-family detached home areas, on the other hand, are said to be less conducive to these kind of suspicious gatherings, and are also said to be carefully watched by property owners concerned with protecting their investments (Tremblay and

32

Dillman, 1983: 39-40). Multi-family attached rental housing is also claimed to be associated with depression and mental illness, while detached housing is held to create a more psychologically liberating environment characterized by open, defensible space (Tremblay and Dillman, 1983:39-40).

33

CHAPTER III QUALITY OF LIFE

In recent years, with the advent of computer technology possessing the requisite memory and processor speed, planners and social scientists have had the capacity to create quantitative indicators or benchmarks of various geographic phenomena over greatly reduced geographic scales and covering the extent of entire cities. Geographic information systems, as well as improved database, spreadsheet and statistical packages have made analysis of great amounts of data more convenient and efficient. Given this new technological capacity, much time and effort has been expended to codify a theory of social and geographic indicators, particularly as relates to the quality of life in urban neighborhoods. The first major influence on quality of life indicator theory comes from the general literature on the construction of social indicators. For data that ranges over a larger geographic scale, such as city-wide data, social indicators began to be used with increasing frequency following the publication of Raymond Bauer's Social Indicators in 1966. Bauer's work was a response to the "philistinism" of conventional economic thinking which equated "quality of life" with monetary "standard of living" measures. Bauer's view was that economic indicators should be distinguished from social indicators, the latter of which gave perspective to the former by showing how standard of living benefits were distributed throughout the community (Bauer, 1966; Marcuse, 1971). The

34

rationale for this was rather simple; if in a given area a subgroup of the population, distributed over the area, was not partaking equally in the economic bounty of that area, social indicators could provide contrast to even the rosiest general economic indicator for that area. Another influence on the development of quality of life indicators came from the Scottish ecological planner Ian McHarg. In his classic work Design With Nature, McHarg introduced urban planners and landscape architects to a perspective which incorporated geographic systems thinking into urban design (McHarg, 1969). The main thrust of McHarg's approach was to use data about geographic regions and ecological systems as a basis for making development decisions. McHarg advised planners and landscape architects to code land according to the part the land played in fundamental environmental processes; if the land was necessarily part of an important environmental process and development of the land would disrupt that process, the land was coded as less suitable for development. McHarg's technique for coding land use is referred to as the "overlay technique," and was an inspiration both for modern GIS practice (Talen, 1997) and modern landscape architecture practice (Leccese, 1997; Spirn, 1984). Such coding of land according to its suitability for development inspired the creation of environmental indicators. Many studies incorporating analysis of such indicator measures into an analysis of environmental systems were undertaken in the ensuing three decades, the most famous of which was the 1972 report by the "Club of

35

Rome" called The Limits To Growth. This report portrayed the vital ecological systems of the entire planet as being fundamentally threatened by the rise in world population and consumption of natural resources (Meadows, Meadows, Randers and Behrens, 1972). Though characterized as "Neo-Malthusian" and unduly pessimistic, the report was the first comprehensive study to measure the systematic effects of the depletion of resources on environmental systems. Following the lead of studies like the "Club of Rome" report, environmental indicator studies have focused on developing measures of the effects of human actions on an environmental system's resource underpinnings, or the material resources necessary to maintain the systems equilibrium (Brown, 1981; Jacobs, 1984). One example of a resource underpinning might be the amount of plant life in an area - since plants play a vital role in the conversion of carbon dioxide to oxygen needed by humans, plants constitute a resource underpinning of the environmental system. The relationship of environmental indicators to economic indicators is somewhat opposite to the relationship of social indicators to economic indicators. Where social indicators look at subpopulations within an area to see if economic benefits are distributed equally, environmental indicators look at the effect of economic policies outside an area to see if economic benefits to an area are predicated on causing harm in a larger geographic context. On the face of it, it would seem that these three groups of indicators form a natural hierarchy, with social indicators measuring phenomena at lower

36

levels of geographic scale, economic indicators measuring phenomena at a mid-range scale, and environmental indicators measuring phenomena at higher scales. The possibility of an "inherent" relationship (Sayer, 1992) between the "three E's" and levels of geographic scale exists to be tested. Within the past two decades, a paradigm has arisen within planning and the social sciences which has turned some of the preceding observations into a general theory of "sustainable development." This paradigm, following the contributions of Bauer and McHarg, sets a standard for an area's "sustainability" which references all three of the above kinds of indicator measures. The sustainability paradigm refers to these indicator groups as the "three E's" (economics, social equity and the environment). If a given area is lacking according to any one of these three sets of indicators, it is considered to be "unsustainable," as even if it scores high on indicators of economic well-being, this wellbeing is either not spread equitably to other segments of the community (as measured by the social indicators), or is not protecting the resource underpinnings by which this wellbeing is produced in the first place (Center for Sustainable Communities, 1997). The past decade has seen the sustainability concept adopted as a theoretical frame for indicator studies undertaken by a number of neighborhood groups and citizen organizations (Jacksonville Community Council, Incorporated, 1997 [1996]; Sustainable Seattle, 1997 [1994.]) The "three E's" model predominates in most of the current spate of

37

community group indicator studies, though other categories of indicators, such as education indicators or health indicators, appeared in these studies from time to time. As the greater capacity for neighborhood-level data analysis became available, a debate emerged in the literature concerning the relative merits of "place prosperity" and "people prosperity" measures (Bolton, 1992; Sawicki and Flynn, 1996). An indicator measure can either represent a geographic entity or a population of individuals, and researchers must concern themselves with whether indicators are constructed with a policy role in mind; since policy is usually administered through geographic areas, indicators of "place prosperity" may be more relevant in terms of policy implementation. By contrast, "people prosperity" is more substantively relevant - places do not enjoy a quality of life, the people who live in them do. Designing written guidelines for indicator measures has been the task of a number of scholars and community activists within the last five years. Major concerns of the writers are that indicators will not be used or considered relevant by the community they are designed to serve (Hart, 1995; Sustainable Seattle, 1997 [1994]) or that they will misrepresent the relationship between that which is measured and the likely urban and regional outcomes which would result from a given indicator score (Luger, 1996). These two considerations, that of community acceptability and relation to likely outcomes, have guided the selection of indicators for most of the major indicator studies undertaken by groups and institutions in the past decade.

38

The sustainability paradigm has provided a workable general theory of indicators or benchmarks. It incorporates some of the insights of public economics, which has long sought a method for measuring social benefits by econometric measures (Becker, 1995) as well as the insights of environmental systems and world-system theorists which stress a focus on an area's context in larger geographic realities (Hopkins, 1982; Straussfogel, 1997; Wallerstein, 1991). The work of Brian Berry and others in establishing a factorial ecology of urban systems (Berry and Horton, 1970; Berry and Kasarda, 1977) has, however, suggested an important refinement to the existing sustainability approach. Through the process of factor analysis (Lorr, 1983; Rummel, 1970), indicator variables may be grouped together in order to discover the "latent structure" of urban systems on the basis of common correlations over a number of areas. The sustainability approach presumes a certain kind of natural grouping of indicators according to function of representing one of the "three E's"; whereas this may be proper, it is also possible that other more latent reasons for indicator grouping exist, and these might tell us more about the nature of the underlying urban system.

39

CHAPTER IV RESEARCH HYPOTHESES

The design of this project will be based around three central hypotheses, which will attempt to evaluate the research question identified in the first chapter of this thesis, drawing upon the existing home ownership and quality of life indicator literatures which have been summarized in the second chapter of the thesis. Each of the three hypotheses will now be set out, with its justification as part of the research design and with an explanation of the specific test to be applied either to verify or falsify it as a substantive claim.

Hypothesis One: Lack of Correlation Between Home Ownership and Quality of Life Indicators H0 - There should be, in general, no strong relationship, positive or negative, between the percentage of home ownership in an area and indicators of quality of life in those areas. Operationally, the definition of a strong relationship for the purposes of evaluating this hypothesis will be over .3 for the value of r2. Thus, correlations between percentage home owners and specific indicator values will not reflect a value of r2 $ .3. The benefit home ownership has for quality of life is presumed by much of the home ownership literature; the contention of this hypothesis is that increased levels of

40

home ownership do not confer any benefit, and thus there will be no strong correlative relationship between home ownership and any of the indicators. If home ownership has been elevated to the status of an "ideology" or a cultural "myth," then one might expect to find a gap between the presumed benefits of a "sacred cow" home ownership and those benefits actually measured in fact; the existence of such a gap is what the preceding hypothesis is designed to establish. The countervailing alternative hypothesis, to be evaluated for each variable at each of the three levels of spatial aggregation, is the following: H1 - There exists a strong positive or negative relationship (an r2 value over .3) between the percentage of home ownership in an area and indicators of quality of life in those areas.

Hypothesis Two: Factor Analysis Of The Indicators Will Produce "The Three E's" H2 - Factor analysis of the indicators will group the indicators into three principal groups, which may be considered as "economic indicators," "social equity indicators" and "environmental indicators." Operationally, a factor will be confirmed from the factor analysis if it has an eigenvalue of 1, and the factors will be orthogonally rotated using the varimax criterion for rotation. The process of factor analysis (which will be described in greater detail in the chapter dealing with this project's statistical methodology) identifies groups of variables with common variance characteristics. By doing this, a number of "factors," or variables which collapse information about other variables with common variance characteristics, may be fashioned. As shown by Brian Berry in his work on factorial ecology, utilizing

41

factor analysis to group indicator variables can be a primary method of identifying underlying "latent" factors in urban ecology systems. The reasoning behind this hypothesis is that the sustainability paradigm's assertion that indicators can be properly grouped into the categories suggested by the "three E's" scheme has never been adequately tested. It is possible that other "latent" factors may be described by the indicator variables. Likewise, it is possible that some or all of the "three E's" might not actually exist as a viable "factor" in the explanation of the quality of life of urban neighborhoods. Should the former proposition be true, the "latent" factors will have to be identified, in keeping with the Berry approach. Should the latter proposition be true, the sustainability paradigm's "three E's" understanding of urban quality of life, at least for the study area of the city of New Orleans, would have to be questioned to a considerable extent. Factor analysis offers a number of criteria by which a factor may be judged as a "true" factor. One of these is the "eigenvalue-1" criterion, which this hypothesis uses. This will be explained in greater detail in the chapter on statistical methodology, but for present purposes it will suffice to say that an eigenvalue is a number which signifies in factor analysis the number of variables which a factor represents. Thus, a factor with an eigenvalue of exactly 1 represents exactly one of the original indicator variables. The expectation is that only three factors will have eigenvalues of 1 or over, and these three factors will correspond to the expected "three E's" of sustainability theory.

42

The rotation of factors is also necessary for improved interpretability of the results of factor analysis; the most commonly-used technique for factor rotation is the "varimax" technique, which is employed for the purposes of evaluating this hypothesis. Once again, more will be said about this in the later chapter, but for present purposes, suffice it to say that varimax rotation simplifies the interpretability of factors by increasing the number of factors which have high correlations with specific variables and decreasing the number of factors which have low or middling correlations with every variable. A factor analysis employing varimax rotation can, thus, better identify factors that represent groups of specific variables, which is the object of this hypothesis. The corresponding null hypothesis for this hypothesis can be stated as follows: H0 - Factor analysis will fail to identify groupings of indicators corresponding to the categories of "economic indicators," "social equity indicators" and "environmental indicators" and will produce groupings of indicators which correspond to a different categorization scheme.

Hypothesis Three: Differential Correlation of Indicator Groups With Home Ownership At Different Geographic Scales H3 - "Social equity" indicators will correlate most strongly with percentage of home ownership at the census block group level. "Economic" indicators will correlate most strongly with percentage of home ownership at the census tract level. "Environmental" indicators will correlate most strongly with home ownership at the neighborhood level. This third hypothesis seeks to establish whether there is indeed an "inherent" relationship between the "three E's" of sustainability theory and the geographic scales

43

which seem to be logically connected to them. If it is true that "social indicators" try to break down considerations of an area's quality of life and measure the quality of life of subareas and subpopulations, and it is also true that "environmental indicators" place an area in its larger context among wider areas and larger populations, then it is likely that "social indicators" will have a stronger explanatory relationship with variables at lower levels of spatial aggregation, while "environmental indicators" will have a stronger explanatory relationship with variables at higher levels of spatial aggregation. Naturally, if the presumed "three E's" factors do not emerge from the factor analysis, a full evaluation of this hypothesis will not be possible; however, it will still be of considerable theoretical importance whether there is some difference in indicator group correlation with percentage home ownership at different geographic scales. The possibility also exists that, at a smaller geographic scale, the more probable homogeneity of residents in an area might affect the correlations of home ownership with the indicators, while at larger scales the likelier diversity of residents may affect it in a different way. The corresponding null hypothesis for this third hypothesis would be stated as follows: H0 - There will be no difference in the correlation levels for specific groupings of indicators corresponding to the differences expected at the three spatial levels of

44

aggregation. Taken together, the three hypotheses venture into the following three important areas of inquiry: 1 - Is there in fact an important and positive relationship between home ownership and indicators of quality of life in urban neighborhoods? 2 - Do indicators of quality of life group into domains of indicators like those suggested by writers in the sustainability paradigm? 3 - Are certain domains of indicators more positively affected by home ownership at different geographic scales?

45

CHAPTER V DATABASE METHODOLOGY

Constructing the Relational Database In order to evaluate the research question, a relational database comprising statistical datasets concerning level of home ownership and quality of life in the city of New Orleans, Louisiana were built and organized spatially at three geographic scales, that of the census block group, census tract and city neighborhood. Information for census block group and census tract areas come from 1990 U. S. Census STF1A demographic files (U. S. Census, 1997 [1990]), and 1994 Topologically Integrated Geographic Encoding and Referencing (TIGER) files from the U. S. Census were used for mapping purposes to represent the boundaries for these census areas (U. S. Census, 1997 [1990.]) Census block groups aggregate up into census tracts with no overlapping boundaries, so a census block group may properly be considered a subdivision of a census tract. The boundaries of city neighborhoods are defined by a document produced by the city of New Orleans during the mayoral administration of Ernest "Dutch" Morial called Neighborhood Characteristics, 1983 (Data Analysis Unit, 1983). This document, on the basis of responses from citizen surveys and an analysis of tract level census data, sets out and defines the boundaries of "the 72 neighborhoods of New Orleans."

46

These neighborhoods are aggregated up, in all but the case of two of the neighborhoods, directly from census tracts which existed in the 1980 census. It was possible to use the equivalent 1990 census tracts to construct these same neighborhood areas without straying significantly from the study's stated criteria for defining the areas it defined as neighborhoods. Such criteria included the identification of areas by their residents as having a certain neighborhood name, general socioeconomic homogeneity, the absence of physical obstacles, identification of neighborhood boundaries by residents, conformity to existing census tract boundaries. Two of the neighborhoods, namely, the Bywater and St. Claude neighborhoods, were composed in part of one census tract that had been split between the two neighborhoods. The rationale for splitting the tract was that St. Claude Avenue is considered by residents to be the firm boundary between the Bywater and St. Claude neighborhoods. The city data analysts elected to observe the importance of that criterion rather than observe the importance of the criterion regarding conformity to existing tract boundaries. This posed a problem in the research process, as the data for neighborhoods needed to be summed from the tract level data in many cases, and for purposes of clarity as well as convenience, it would be preferable to have neighborhoods be aggregated up from census tracts in every case. Since conformity to census tract boundaries was one of the city's criteria in the creation of the original 72 neighborhoods, the liberty was taken of

47

adjusting the boundaries of the Bywater and St. Claude neighborhoods to be coterminous with census tracts. The entirety of the split census tract was assigned to the Bywater neighborhood, as a majority of the tract lies south of St. Claude Avenue. Otherwise, the neighborhoods are composed of exactly the tracts the city's Data Analysis Unit assigned to them in 1983. Three datasets in the relational database correspond to the three levels of spatial aggregation: the block group level, the tract level and the neighborhood level.
Table 4. Number of Cases in Geographic Datasets Geographic Scale Level Number of Cases Block Group 657 Tract 184 Neighborhood 72

The data matrix in each dataset ranges over 38 variables, which represent quality of life indicator variables, which will be defined later in this chapter. Thus, the data matrix for the block group dataset is a 657 x 38 matrix, the data matrix at the tract level is a 184 x 38 matrix, and the data matrix at the neighborhood level is a 72 x 38 matrix. Additionally, each dataset contains a variable representing the percentage of home ownership to be compared to the indicator values in the data matrix. There were four kinds of indicator variables constructed for this project: U. S. Census sociodemographic variables, point-in-polygon spatial join variables, centroid distance variables and perimeter distance variables. The indicator variables are listed on the following page with their method of construction identified (Table 5). Each of these

48

Table 5. Indicator Variables Identified By Variable Construction Type Variable Name Class Hom eownership Census Ratio Bicycling To W ork Census Ratio CAT Elem entary Test 50th Percentile+ Centroid Distance Clothing Stores Centroid Distance Com m uting Index Census Ratio Diversity of Em ploym ent Census Ratio Doctors Centroid Distance Doctors Point-In-Polygon Join Em ploym ent Census Ratio Fire Stations Centroid Distance Graduate Exit Exam 90%+ Centroid Distance Grocery Stores Centroid Distance Grocery Stores Point-In-Polygon Join Health Risk Centroid Distance High School Graduation Rate Census Ratio Hom e Affordability Census Ratio Hospitals Centroid Distance LEAP Elem entary Test 90%+ Centroid Distance LEAP Jr./Middle Test 90%+ Centroid Distance Libraries Centroid Distance Median Household Incom e Census Ratio Median Property Value Census Ratio Nonviolent Crim e Point-In-Polygon Join Parks Centroid Distance Parks Perim eter Distance Police Stations Centroid Distance Post-High School Graduation Rate Census Ratio Public Assistance Census Ratio Real Em ploym ent Census Ratio Recreation Facilities Centroid Distance RTA To W ork Census Ratio Schools Centroid Distance Toxic Releases Centroid Distance Toxic W ater Releases Centroid Distance Traffic Index Census Ratio Vacancies Census Ratio Violent Crim e Point-In-Polygon Join Violent Non-Property Crim e Point-In-Polygon Join W alking To W ork Census Ratio

Treatment None None None None None None None Norm alized None None None None Norm alized None None None None None None None None None Norm alized None None None None None None None None None None None Norm alized None Norm alized Norm alized None

By Area

By Area

By Area

By Area By Area By Area

49

four kinds of variables were produced through the use of ArcView 3.0a geographic information systems software, exported for the purpose of organization as datasets (to the Microsoft Excel spreadsheet and Microsoft Access database applications) and eventually transferred as complete datasets for analysis (to the Statistical Package for the Social Sciences for the Personal Computer, or SPSS-PC, Version 7.0 statistical analysis application). The four methods for variable construction each address a different goal of geographic modeling of quality of life. A general description of the methods can be provided at this point: · U. S. Census sociodemographic variables. The simplest measures used in this project database are directly taken from the STF1A demographic files of the 1990 United States Census, representative variables for which were downloaded from the "Census Lookup" World Wide Web site (U. S. Census, 1997 [1990.]) Census information exists on this web server at the block group and tract level of aggregation; all neighborhood level data for sociodemographic indicator variables were fashioned by aggregating up from the tract level data. The sociodemographic variables provide a percentage measurement, based on the ratio of two or more census variables. · Point-in-polygon spatial join variables. ArcView 3.0a has the capacity to perform what is called a "spatial join" of geographic shape coverages. A

50

"spatial join" takes database information for one of the themes and combines it with database information for the other theme, on the basis of some common geographic characteristic. To take an example, if one has a coverage representing a group of points and a coverage representing a group of polygons, and one wishes to know what polygon each of the points in the point coverage are in, the "spatial join" feature of ArcView can provide this information in tabular form. In other words each, point of the point coverage, after the procedure is finished, will be listed along with the specific polygon that point is in. This describes the "point-in-polygon" spatial join procedure. The utility of this procedure for the construction of indicator variables lies in the procedure's capacity to provide information about the frequency of points within each polygon area; once the data has been set up in tabular format, frequency statistics may be run to determine how many points are in each polygonal area. Since, in this project, the polygonal areas are block groups, tracts and neighborhoods, point-in-polygon frequencies reflect the number of points within each of these kinds of polygons. Presuming the points represent features which affect quality of life in those areas, the point-in-polygon spatial join provides information on how many of these features there are in each block group, tract or neighborhood. Each of these variables is normalized by area, to control for the misrepresentation of point frequency data which would

51

occur if one did nothing to account for differences in the size of the polygons in which those points are to be found. Large polygons simply by virtue of their size have room for more points, and it is more meaningful to measure the number of points per square mile than to simply measure the number of points per polygon. · Centroid distance variables. These variables represent the distance of a point feature from the centroid, or geographic center, of a polygon. The utility of this method of indicator construction lies in its capacity to measure a distance from a point, which would represent a feature which affects quality of life to the center of a polygonal area of interest, which would represent a block group, tract or neighborhood. A program script written in the Avenue macro language which comes with the ArcView 3.0a software package made it possible to create point coverages representing all the centroids for the block groups, tracts and neighborhoods; centroid distance variables are calculated using another Avenue script which create "spider diagrams" that measure the distance from the nearest of the point features in a point coverage to these geographic center points of block groups, tracts and neighborhoods. · Perimeter distance variables. This kind of variable represents the distance of a polygon centroid from the nearest part of the outer perimeter or edge of another polygon. The assumption behind this method of indicator

52

construction is that it is more important to measure how far away from the edge of a large area one is than how far away from the center of that area. There is only one variable in the project database which has been constructed by this method, and its object was to measure the distance from block groups, tracts or neighborhoods to city parks. A separate measure, calculated by the centroid distance method, measures the distance from the center of a block group, tract or neighborhood to the nearest center of a city park. However, it may be that, especially in the case of large parks, how close one is to the edge of the park is of more concern than how close one is to the park center. This method of indicator construction meets that objection. Using the ArcView 3.0a "Select By Theme" feature, it was possible to get values for the distance of block group, tract and neighborhood centroids to the perimeter of the park polygons, and these values were entered into the database for each of the block groups, tracts and neighborhoods. The data for the indicator variables came from a variety of sources. Table 6 identifies these sources and the group of indicators which came from each source. The majority of the variables are constructed from information in the U. S. Census. Other sources include maps of the city of New Orleans (in particular the Gousha and RandMcNally city "street finder" maps), the New Orleans Fire Department (NOFD), the New Orleans Police Department (NOPD) with assistance from the Metropolitan Crime

53

Table 6. Sources for Indicator Variables Name Hom eownership Bicycling To W ork Com m uting Index Diversity of Em ploym ent Em ploym ent High School Graduation Rate Hom e Affordability Median Household Incom e Median Property Value Post-High School Graduation Rate Public Assistance Real Em ploym ent RTA To W ork Traffic Index Vacancies W alking To W ork Parks Parks Fire Stations Nonviolent Crim e Violent Crim e Violent Non-Property Crim e Police Stations Clothing Stores Doctors Doctors Grocery Stores Grocery Stores Hospitals Libraries Recreation Facilities Schools CAT Elem entary Test 50th Percentile+ Graduate Exit Exam 90%+ LEAP Elem entary Test 90%+ LEAP Jr./Middle Test 90%+ Health Risk Toxic Releases Toxic W ater Releases

Source Census Census Census Census Census Census Census Census Census Census Census Census Census Census Census Census City Map City Map NOFD NOPD NOPD NOPD NOPD/MCC Switchboard Switchboard Switchboard Switchboard Switchboard Switchboard Switchboard Switchboard Switchboard Tim es-Picayune Tim es-Picayune Tim es-Picayune Tim es-Picayune TRI TRI TRI

54

Commission (MCC) of New Orleans, the "Switchboard" telephone directory on the World Wide Web, the New Orleans Times-Picayune and the United States Environmental Protection Agency Toxic Release Inventory (United States Environmental Protection Agency, 1997 [1995]).

Indicator Construction Considerations In the process of building the indicator database, a number of issues came up which seem to bear importantly on the theory of indicator design. These issues, along with the resolutions deemed appropriate for the issues, will be discussed in this portion of the thesis. One major question which seems not to have been dealt with in the literature is what to do about spatial obstacles in the calculation of distance measurements. Distance measurements to important locations are often used as indicator measures; one typical interpretation of a distance indicator is that "spatial mismatch" exists in a certain area where people live far from important locations, such as their place of employment (Luger, 1996). Such indicators gauge the proximity to features; the closer to the feature, the more positive the value of the indicator. However, the proximity of a location does not always imply easy access to the location. There may be obstacles along a short path to a feature, while a longer path is clear of obstacles.

55

The following map (Figure 3) provides an example of how spatial obstacles may intervene in the construction of a distance indicator. The indicator variable called "Libraries" is a measure of the distance to library buildings from polygon centroids. In the map shown, the distances are from the libraries to block group centroids; each of the lines radiating out from the library locations represents the distance from a block group centroid to the nearest library. The difficulty with spatial obstacles can be immediately seen graphically on the map - in some cases, the nearest library to a particular block group centroid is across the Mississippi River from that block group. Most people would hold that the presence of a river presents a formidable obstacle to travel to a nearby library, and thus, it may be argued that merely measuring the proximity of libraries, without adjusting the figure somehow to reflect spatial obstacles in one's path, is misleading. It is possible that a "second farthest" criterion for distances when the closest path traverses the river should have been adopted, such that all distances would reflect the shortest path to a library that does not require crossing the river. However, such a criterion would not solve this problem in all cases. There is, for example, a bridge crossing the river at one point, and for many the availability of the use of the bridge would solve the spatial obstacle problem. Yet one needs to have an automobile to cross this bridge, and those who do not have one may still consider the river an obstacle and prefer to use a library farther from

56

Figure 3. Block Group Centroid Proximity to Libraries

57

them to which they have a clearer path. Another complication is that there is ferry and bus service across the river which could get a person on one side of the river to a nearby library on the other side. Those who find the ferry or bus service dependable might cross the river to get to the closer library, but those who do not might go to the one farther away instead. The practical upshot of all of this is that controlling for the spatial obstacle problem is an extremely complicated proposition. Indeed, even after considering the above matters, rivers are not the only kind of spatial obstacle, and the definition of what can be properly considered an obstacle may change from person to person. A bicyclist, for example, might find crossing the New Orleans Industrial Canal to be extremely difficult, and therefore consider the canal a spatial obstacle. On the other hand, most car drivers in the New Orleans East area cross it daily and consider it nothing of the sort. Also, many in the city would voluntarily take a longer trip to avoid high-crime areas in the city, which for them would therefore constitute a sort of spatial obstacle; others would have no qualms about going through these areas and would deny they are any kind of obstacle. Another consideration is that proximity is measured in this analysis “as the crow flies,” and the actual road distance or travel time to the selected destinations may complicate decisions greatly. The question which must be raised here is whether any kind of spatial obstacle necessarily takes precedence in importance over others. The decision was made for the

58

purposes of this analysis to make no adjustments to reflect the presence of spatial obstacles, but this does not mean that others may not wish to attempt this, complicated though the task may be. The reason no such adjustment was undertaken, besides the sheer complexity of the task, is that allowing the crossing of spatial obstacles in indicator calculations allows for a moderate "benefit of the doubt." In other words, if the closest library to a block group is across the Mississippi River from that block group, one thing is certain - there is no library closer to the library than at least the width of the river. Thus we can assume that the next farthest library away is, indeed, fairly far away. So using the nearest library across the river provides an optimistic estimate for the indicator, but one which will still be low, and which would be even lower if the next farthest library were used to calculate the distance. Since any error involved with leaving the calculation as it stands would be either moderate, in the favor of the block group, or both, the choice to do so did not seem particularly unfair, and thus the original calculation was retained. Another concern which raised itself during the database construction phase of this project was the philosophical basis for valuing proximity in distance calculations. The well-known urban ethic called by its detractors "NIMBY" (the acronym stands for "Not In My Back Yard") may be important to understand contextually when dealing with indicator measures involving distance. It may be, for example, possible that something that people want close to where they live may not be something they want where they live. A term one might coin for this would be "NIMBY-BANJO" ("Not In My Back

59

Yard - But Another Neighborhood Just Over.") A good example of this sort of indicator might be the one constructed for this project called "Hospitals." This indicator measures the distance from area centroids to the nearest hospital. Hospitals are, of course, crucially important to have near where one lives. On the other hand, due to the high intensity of use which occurs at hospital sites, coupled with parking and noise problems, hospitals might often be considered by area residents to be locally unwanted land uses (often referred to as "LULUs.") For this reason, proximity to a hospital would be a clear "NIMBY-BANJO" kind of measure; closeness (but not too much closeness) to a hospital is a measure of quality of life. The problem with "NIMBY-BANJO" measures is similar to the problem earlier identified with spatial obstacles. If one makes an effort to compensate for this problem, what form should the compensation take? Since people generally want to be both somewhat far from hospitals and generally close to them, it would seem logical to create some kind of curve measure, and at some point on the curve proximity to a hospital would stop negatively affecting the indicator and start to influence it positively. Of course, the problem with that is specifying at what point proximity to a hospital is bad and at what point it is good. One thing remains clear, and that is that it is vitally important to be at least reasonably close to a hospital, and whenever a measure can be justly called a "NIMBYBANJO" measure, there must be some similar reason why it is important that something

60

at least be in "another neighborhood just over." For the purposes of this thesis, all "NIMBY-BANJO" measures were pure measures of proximity; the importance of the proximity was considered to be overriding. The idea that this conferred "the benefit of the doubt" upon an area was once again a rationale for the decision; even if one felt that a hospital in one's immediate neighborhood somewhat lowered the quality of life, at least one would be assured there is a nearby hospital in case of a medical emergency, which would somewhat raise the quality of life. Once again, the decision not to adjust "NIMBY-BANJO" variables does not mean that others may not wish to develop means to do so.

Indicator Justification Some elaboration of the philosophical bases for indicator selection also seemed necessary before selecting the indicators themselves; the quality of life indicator literature shows that the selection of an indicator usually depends on the indicator's acceptability to a given community or the relation of that indicator to likely urban or regional outcomes. In the selection process, some rules of indicator justification were created to support the choice to include specific quality of life indicators in the database. The first objective of establishing rules for justification of the indicators was to somehow rate indicators on the basis of their likely community acceptability. Since no

61

attempt to survey citizens of New Orleans neighborhoods was attempted in the course of this project, another way of rating this was needed. One solution to the problem presented itself in the notion of setting up a rule of thumb which tests an indicator on the basis of the likely clear rejection by the community of the opposite of what the indicator measures. This might be called the "rule of unthinkability." The rule could be formally stated as follows: An indicator is better justified as a measure of quality of life if the opposite of what the indicator measures is something a community would clearly find to be unthinkable and worthy of rejection on its face as a measure of quality of life. The idea behind this is that no quality of life indicators are reasonable if one might as well make measuring the exact opposite an indicator of quality of life. An example of this might be making population growth an indicator of quality of life. There are, indeed, many who would so consider population growth; the more people who arrive in a city, the higher the quality of life in the city must be to attract them. However, those living in high-growth areas such as Florida or Seattle, Washington might not agree with the idea that population growth is a measure of quality of life, and some might suggest quite the opposite - that the less population growth there is, the better the quality of life will be. In this case, the opposite of what is being measured is not only not "unthinkable," it may even be strongly advocated by some as a quality of life measure in its own right.

62

The "rule of unthinkability" would prevent both sides from establishing their measures, as there is clearly a flip side to the coin. One person's hell is another person's heaven, and vice-versa; in such circumstances, no indicator will avoid contention and the likelihood of community acceptability will be considerably lessened. A broad extension of the "rule of unthinkability" would make indicator measures even less likely to be disputed as measures. For example, there may be certain communities where, one may suppose, the number of individuals who wear three-piece suits may be considered a valid indicator measure. If, in that community, it is considered "unthinkable" not to wear three-piece suits, then the indicator measure would certainly stand muster in that community. However, a neighboring community may find the number of individuals who do not wear three-piece suits to be a valid indicator measure. Thus the "unthinkable" is thinkable, and the indicator measure fails the test. Considered broadly, the "rule of unthinkability" would exclude any indicator measures on this threepiece suit subject; the object of the rule is to increase the likelihood of an indicator's general acceptance, and a broad consideration of the rule makes it clear that the indicator would not be generally accepted. Though the adoption of such a rule does not entirely avoid the charge that indicator measures are somewhat subjective, it does establish a high standard for the intersubjectivity necessary to adopt an indicator. Gaining a greater objectivity would be the second objective of establishing rules of indicator justification. The rule of thumb which seems to best address this concern is

63

what might be called the "rule of necessity," which would test an indicator on the basis of its presumed relationship to the necessities of human existence. Stated formally, the rule would look like this: An indicator is better justified if it bears a clear and objectively supportable short-term or long-term relationship to the maintenance and protection of human life. Such measures would literally be measures of the quality of human life per se. Naturally, that which maintains and protects human life is also often that of which the opposite is "unthinkable." But this gets more to the concern that quality of life indicators ought to relate to urban and regional outcomes. Regardless of the subjective opinions of individuals about what makes for a good quality of life, there are undeniable and objective concerns, like the need for food, clothing, shelter, education, health and safety. Indicators which measure the prospects for the maintenance and protection of human life are thus clearly and strongly justified. Short-term indicators of this type make a better case, as long-term indicators of necessity may be criticized as unduly "pessimistic" by those who think the long term problem may be alleviated before it comes to fruition; this is, in fact, a common criticism of environmental indicators (Kahn, Brown and Martel, 1976). However, long-term indicators are not necessarily invalid because they are long-term, if their premises are supportable by evidence.

64

The Indicators Now that the problems with indicator construction and the rationales for indicator choice have been set out, each of the indicators themselves may now be presented in depth. With each of the indicator variables a short description is provided, as well as the data source for the variable, a brief explanation of how the variable was calculated, whether a positive or negative score on the variable indicates a favorable quality of life, and the justification for choosing the measure in the first place. If an indicator measure is marginally inaccurate, because it reflects the influence of spatial obstacles or is characterized by the high intensity of use which is part and parcel of the "NIMBYBANJO" problem, information concerning this is also provided. In the following, an "area" refers to a block group, tract or neighborhood. References to the "rule of unthinkability" will be prefaced by "U:" and references to the "rule of necessity" will be prefaced by "N:" in order to provide a clearer exposition of the logical premises involved in the construction of each indicator. Indicator 1. Bicycling To Work Description: The percentage of the total population in an area who reported regularly bicycling as a means of transportation to a place of employment. Source: 1990 United States Census. Calculation: The number of persons bicycling to work / The total population of the area. Value Should Be: Larger. Justification: This indicator measures a number of things which bear greatly on urban quality of life. It serves to measure whether streets in an area are safe on which to ride bikes. (U: It would be unthinkable to have unsafe streets). It also measures the possibility of reaching one's place of employment by means of a

65

vehicle that can only go short distances, and thus serves as a proxy measure for proximity to one's place of employment. (U: It would be unthinkable to require people to live extremely far from where they work; N: It could cause economic hardship to require people to live extremely far from where they work, as suggested by the "spatial mismatch" literature). Finally, bicycle riding conserves fossil fuels and lowers automobile exhaust emissions. (N: Fossil fuels are necessary for economic production and public services and thus should be conserved; N: Automobile exhaust increases the likelihood of air pollution, which threatens human health). Spatial Obstacles? No. This is actually in part a measure of the lack of spatial obstacles, so such obstacles do not affect the measure's accuracy. High-Intensity Uses? Yes. One may not wish to live right next to where one works, but it is important for economic reasons as well as reasons of convenience to live reasonably close to employment. Indicator 2. CAT Elementary Test 50th Percentile and Above Description: The distance from the centroid of an area to the nearest public elementary school at which students scored in the 50th percentile or higher on the California Achievement Test. Source: New Orleans Times-Picayune. Calculation: Calculated via an ArcView Avenue macro language script which identifies the closest point feature in a geographic coverage to a point feature in another coverage. Represents an average of the 4th grade and 6th grade test scores. Value Should Be: Smaller. Justification: This indicator measures the proximity to a publicly-funded school which has achieved a level of educational excellence. Public schools generally serve a specific geographic area. (U: It would be unthinkable to require people to send their children to poor public schools). In the case of "magnet" schools, this indicator would measure the ability of parents to conveniently get their children to the school location every morning. (U: It would be unthinkable to require parents to transport their children extremely long distances each morning to get them to school). Also, if children are expected to walk to school, school ought to be close by. (U: It would be unthinkable to expect children to walk a long distance to school). Since public schools are funded by tax money, which residents of the city must pay, it is important that citizens get what they pay for. (U: It is unthinkable that persons be charged tax money for inferior social services). Finally, education increases one's productivity and economic prospects. (N: Education increases one's economic chances in life).

66

Spatial Obstacles? Yes. Particularly since this indicator deals with the safety of small children and the logistics of conducting them to school, any spatial barriers might bear strongly on the interpretability of this indicator. High-Intensity Uses? Possibly. Some might find public elementary schools to be a bothersome high-intensity use, but it is probably many communities would find the presence of elementary schools to be appropriate. Indicator 3. Clothing Stores Description: The distance from an area centroid to the nearest retail establishment which is identified as a clothing store. Source: Switchboard telephone directory website. Calculation: Calculated via an ArcView Avenue macro language script which identifies the closest point feature in a geographic coverage to a point feature in another coverage. Value Should Be: Smaller. Justification: Clothing is a necessary item for protection against the elements. (N: Clothing is necessary for human life). Spatial Obstacles? Yes. The existence of spatial barriers probably does not bear greatly on the interpretability of the indicator, however, as small amounts of clothing can be transported larger distances. The indicator can still provide information on whether clothing stores are or are not within a reasonable proximity of an area. High-Intensity Uses? Possibly. Clothing stores are only high-intensity uses if they are large establishments. Low-scale business use may be compatible with a residential area. Indicator 4. Commuting Index Description: This indicator represents how much time it takes for commuting persons who live in an area to get to their place of employment. It is an aggregated measure in which commuting times are grouped into 12 "waves," with the commuters who take the least time to get to work being in the first wave of commuters and the commuters who take the most time to get to work being in the twelfth. Source: 1990 United States Census. Calculation: The census provides the number of individuals in each of the twelve waves. An algorithm was written in Microsoft Excel to produce number of the wave category containing the most members for each area. This number represents the commuting wave the area is in. In the case of an area with two or

67

more categories with the most members, the lowest wave was chosen. In other words, if an area had the most commuters in both the second wave and the fifth wave, the area was considered to be in the second wave of commuters. Thus, any error is on the side of giving the area the "benefit of the doubt." Value Should Be: Smaller. Justification: Long commutes are undesirable in terms of lost time. (N: Lost time could be used for more economically profitable purposes). Increased traffic resulting from commuting can cause jams to occur. (U: It is unthinkable to promote traffic jams). Spatial Obstacles? No. Spatial obstacles could only increase commuting time for some, so any error would be in the direction of giving a "benefit of the doubt" to the area. Since this is a census statistic, the value of the indicator represents the empirical responses of census survey respondents, so the effect of spatial obstacles would have to be considered accounted for. High-Intensity Uses? No. Indicator 5. Diversity of Employment Description: The census provides a group of variables identifying 10 economic sectors in which individuals are employed, with each variable representing the number of persons in an area employed in each of the sectors. The gap between number of persons employed in the largest sector, or sector with the most number of employees in the area, and the smallest sector, or sector with the fewest number of employees in the area, provides a measure of employment diversity in the area. A large gap means that most of the labor force in that area is employed in one industry, whereas a small gap means that the labor force is more diversified and is spread out over many industries. Source: 1990 United States Census. Calculation: An algorithm was calculated in Microsoft Excel to find the maximum value for the number of persons in the economic sectors for each area and the minimum value for the economic sectors for each area. The value of the indicator = maximum value - minimum value. Value Should Be: Smaller. Justification: An area which is not employed entirely in one industry will stand a better chance of not collapsing economically if that industry declines. (N: Economic diversification gives an area a better chance of survival). Spatial Obstacles? No. High-Intensity Uses? No.

68

Indicator 6. Doctors (Centroid Distance) Description: The distance from an area centroid to the nearest doctor's office. Source: Switchboard telephone directory website. Calculation: Calculated via an ArcView Avenue macro language script which identifies the closest point feature in a geographic coverage to a point feature in another coverage. Value Should Be: Smaller. Justification: Proximity to trained, on-duty physicians, particularly in the case of an emergency, is important for human health. (N: Proximity to doctors is important for human health). Spatial Obstacles? Minimally. Though the centroid distance measures often have this problem, there are enough doctors in New Orleans for it not to be terribly difficult to get to one in most areas. This, indeed, prompted the calculation of the second version of the "doctor indicator" which measures the number of doctor's offices in an area. However, there are areas which are still somewhat far removed from medical services, and spatial obstacles would make the evidently long trips to the doctor even longer. High-Intensity Uses? Possibly. Doctors may easily locate their practices in residential neighborhoods without raising concerns about the high intensity of land use. Doctors may also, however, work at hospitals, which have occasion a great deal of high-intensity land use. Indicator 7. Doctors (Point-in-Polygon Spatial Join) Description: The number of doctors per square mile located within an area. Source: Switchboard telephone directory website. Calculation: Calculated by joining a geographic point coverage representing doctors with a practice located in the city of New Orleans with geographic polygon coverages representing city block group, tract and neighborhood areas. This yields a data set in which all of the points are coded by the area in which they are found. Then frequency statistics are calculated to get a number for each area corresponding to the number of doctors in each area. This number is then divided by the number of square miles in the area to normalize the figure to represent doctors per square mile in the area. Value Should Be: Larger. Justification: Proximity to trained, on-duty physicians, particularly in the case of an emergency, is important for human health and safety. (N: Proximity to doctors is important for human health and safety). Proximity to a larger number of physicians increases the likelihood of being close to a physician competent in

69

the area of expertise required for a specific ailment. (N: Proximity to medical specialists is important for human health and safety). Spatial Obstacles? No. This measure indicates the number of doctors in the area itself. High-Intensity Uses? Very probably, especially since a larger number of physicians indicates the likely presence of a hospital, and hospitals tend to occasion high-intensity uses. Indicator 8. Employment Description: The percentage of employed members of the area's labor force. Source: 1990 United States Census. Calculation: The number of employed persons / Those defined as members of the labor force by the census. Value Should Be: Larger. Justification: This is a commonly-used indicator of the economic health of a community. Those who are employed have the means to economically support themselves and dependents. (N: Employment is an economic necessity in the absence of inherited income or a benefactor). Spatial Obstacles? No. High-Intensity Uses? No. Indicator 9. Fire Stations Description: The distance from an area centroid to the nearest fire station. Source: New Orleans Fire Department. Calculation: Calculated via an ArcView Avenue macro language script which identifies the closest point feature in a geographic coverage to a point feature in another coverage. Value Should Be: Smaller. Justification: Proximity to a fire station is necessary for physical security and security of property in case of a fire. (N: Fire stations guard against threats to human life and property in the event of a fire). Spatial Obstacles? Yes. However, this does not bear greatly on the interpretability of the measure, as fire trucks can surmount many obstacles and generally travel at great speeds. High-Intensity Uses? Possibly. Fire stations are usually low-intensity of use sites, except when called into action. Those times may be sufficiently jarring and noisy to pose a nuisance to the community, however.

70

Indicator 10. Graduate Exit Exam 90% or More Description: The distance from area centroids the nearest public high school at which 90% of the students or more passed the Graduate Exit Exam administered by the Louisiana Department of Education. Source: New Orleans Times-Picayune. Calculation: Calculated via an ArcView Avenue macro language script which identifies the closest point feature in a geographic coverage to a point feature in another coverage. Represents an average of all of the separately-administered sections of the GEE exam. Value Should Be: Smaller. Justification: This indicator measures the proximity to a publicly-funded school which has achieved a level of educational excellence. Public schools generally serve a specific geographic area. (U: It would be unthinkable to require people to send their children to poor public schools). In the case of "magnet" schools, this indicator would measure the ability of parents to conveniently get their children to the school location every morning. (U: It would be unthinkable to require parents to transport their children extremely long distances each morning to get them to school). Also, if children are expected to walk to school, school ought to be close by. (U: It would be unthinkable to expect children to walk a long distance to school). Since public schools are funded by tax money, which residents of the city must pay, it is important that citizens get what they pay for. (U: It is unthinkable that persons be charged tax money for inferior social services). Finally, education increases one's productivity and economic prospects. (N: Education increases one's economic chances in life). Spatial Obstacles? Yes, though high school students do tend to be more mobile, so obstacles should not bear too greatly upon the interpretability of this measure. High-Intensity Uses? Yes. Public high schools are quite likely to be considered undesirable high-intensity uses. Indicator 11. Grocery Stores (Centroid Distance) Description: The distance from area centroids to the nearest grocery store. Source: Switchboard telephone directory website. Calculation: Calculated via an ArcView Avenue macro language script which identifies the closest point feature in a geographic coverage to a point feature in another coverage. Value Should Be: Smaller. Justification: Food is a necessary item for human existence, and grocery stores are a primary food source for many individuals. (N: Grocery stores are primary

71

providers of necessary food items and hence are important for human survival). Also, proximity to grocery stores is important for elderly members of the population, who may lack adequate transportation as well as physical strength to carry large amounts of groceries. (U: It would be unthinkable to require elderly individuals to travel long distances for their groceries). Spatial Obstacles? Minimally. Though the centroid distance measures often have this problem, there are enough grocery stores in New Orleans for it not to be terribly difficult to get to one in most areas. This, prompted the calculation of the second version of the "grocery indicator" which measures the number of grocery stores in an area. However, there are areas which are still somewhat far removed from grocery stores, and spatial obstacles would make the evidently long trips to the grocer's even longer. High-Intensity Uses? Possibly. "Corner" grocery stores are low-intensity uses, though they may possibly attract loiterers and encourage disruptions. Large-scale groceries, on the other hand, are very much high-intensity uses, and are likely to cause parking problems and traffic congestion. Indicator 12. Grocery Stores (Point-in-Polygon Spatial Join) Description: The number of grocery stores per square mile located within an area. Source: Switchboard telephone directory website. Calculation: Calculated by joining a geographic point coverage representing grocery stores located in the city of New Orleans with geographic polygon coverages representing city block group, tract and neighborhood areas. This yields a data set in which all of the points are coded by the area in which they are found. Then frequency statistics are calculated to get a number for each area corresponding to the number of grocery stores in each area. This number is then divided by the number of square miles in the area to normalize the figure to represent grocery stores per square mile in the area. Value Should Be: Larger. Justification: Food is a necessary item for human existence, and grocery stores are a primary food source for many individuals. (N: Grocery stores are primary providers of necessary food items and hence are important for human survival). Proximity to grocery stores is important for elderly members of the population, who may lack adequate transportation as well as physical strength to carry large amounts of groceries. (U: It would be unthinkable to require elderly individuals to travel long distances for their groceries). In addition, if there are a number of grocery stores within an area, price competition will be more likely to occur, as going to a competing grocery store will be facilitated. (U: It would be

72

unthinkable to not have strong market competition regulate the price of food items). Spatial Obstacles? No. This measure indicates the number of groceries in the area itself. High-Intensity Uses? Yes. A larger number of grocery stores would indicate the area is a commercial area and one would expect such an area to be characterized by high intensity of use. Indicator 13. Health Risk Description: The distance from area centroids to sites where toxic chemicals are discharged into the environment equal to or greater than a ratio of 100 pounds for every 1 part per million (ppm) allowed by the United States Occupational Safety and Health Administration (OSHA) standards. The closer the area centroid is to the toxic site, the greater the implied health risk to the area. Source: Environmental Protection Agency Toxic Release Inventory (TRI) and the United States Occupational Safety and Health Administration. Calculation: The measure of health risk at each site was calculated according to the following formula: Pounds discharged at the toxic release site / OSHA PPM threshold standard. All sites with a value for this ratio of less than 100 were eliminated from consideration for the distance measure. Then, calculated via an ArcView Avenue macro language script which identifies the closest point feature in a geographic coverage to a point feature in another coverage. Value Should Be: Larger. Justification: Toxic chemicals discharged into air, water or land pose a health threat to the immediate environment of an area. (N: Toxic chemicals are potentially harmful to human health). Spatial Obstacles? No. High-Intensity Uses? No. Toxic sites are, of course, generally industrial plants with a tremendously high intensity of use. However, this indicator has a positive implication for quality of life the longer the distance from the site, so there is no concern about this being a "NIMBY-BANJO" indicator; indeed, toxic sites are better described as a wholly unwanted land use - a traditional "NIMBY" or "LULU." Indicator 14. High School Graduation Rate Description: The percentage of persons 18 and over in the area who have graduated high school. Source: 1990 United States Census.

73

Calculation: The number of persons having graduated from high school / The number of persons 18 and over. Value Should Be: Larger. Justification: Education increases one's productivity and economic prospects. (N: Education increases one's economic chances in life).. Spatial Obstacles? No. High-Intensity Uses? No. Indicator 15. Home Affordability Description: The ratio of the median housing value of an area to the median household income of the area. Source: 1990 United States Census. Calculation: Median housing value / Median household income. Value Should Be: Smaller. Justification: Shelter is a necessary item for human existence. A lower price for shelter is economically desirable. (N: Shelter is a necessary item, and the price of a necessity should be as low as possible). Spatial Obstacles? No. High-Intensity Uses? No. Indicator 16. Hospitals Description: The distance from an area centroid to the nearest hospital. Source: Switchboard telephone directory website. Calculation: Calculated via an ArcView Avenue macro language script which identifies the closest point feature in a geographic coverage to a point feature in another coverage. Value Should Be: Smaller. Justification: Hospital emergency rooms and hospital surgery and technical facilities are vital resources in the event of a major ailment or sickness requiring immediate attention. Proximity to a hospital, and particularly to an emergency room, is of paramount importance. (N: Proximity to a hospital is very important for the protection of human life and health). Spatial Obstacles? Yes, though hospitals often have the resources to surmount spatial obstacles in order to get to an individual in an emergency, via ambulence service or even helicopter airlift in some cases. High-Intensity Uses? Yes. Hospitals are extremely high-intensity use sites, and are perhaps the best example in this database of a "NIMBY-BANJO" indicator.

74

Indicator 17. LEAP Elementary Test 90% or More Description: The distance from an area centroid to the nearest elementary school at which 90% or more of the students passed the Louisana Educational Assessment Program test. Source: New Orleans Times-Picayune. Calculation: Calculated via an ArcView Avenue macro language script which identifies the closest point feature in a geographic coverage to a point feature in another coverage. Represents the average of the 3rd grade and 5th grade test scores. Value Should Be: Smaller. Justification: This indicator measures the proximity to a publicly-funded school which has achieved a level of educational excellence. Public schools generally serve a specific geographic area. (U: It would be unthinkable to require people to send their children to poor public schools). In the case of "magnet" schools, this indicator would measure the ability of parents to conveniently get their children to the school location every morning. (U: It would be unthinkable to require parents to transport their children extremely long distances each morning to get them to school). Also, if children are expected to walk to school, school ought to be close by. (U: It would be unthinkable to expect children to walk a long distance to school). Since public schools are funded by tax money, which residents of the city must pay, it is important that citizens get what they pay for. (U: It is unthinkable that persons be charged tax money for inferior social services). Finally, education increases one's productivity and economic prospects. (N: Education increases one's economic chances in life). Spatial Obstacles? Yes. Particularly since this indicator deals with the safety of small children and the logistics of conducting them to school, any spatial barriers might bear strongly on the interpretability of this indicator. High-Intensity Uses? Possibly. Some might find public elementary schools to be a bothersome high-intensity use, but it is probably many communities would find the presence of elementary schools to be appropriate. Indicator 18. LEAP Jr./Middle Test 90% or More Description: The distance from area centroids to the nearest junior high school or middle school at which 90% or more passed the Louisiana Educational Assessment Program test. Source: New Orleans Times-Picayune.

75

Calculation: Calculated via an ArcView Avenue macro language script which identifies the closest point feature in a geographic coverage to a point feature in another coverage. Represents the 7th grade test score only. Value Should Be: Smaller. Justification: This indicator measures the proximity to a publicly-funded school which has achieved a level of educational excellence. Public schools generally serve a specific geographic area. (U: It would be unthinkable to require people to send their children to poor public schools). In the case of "magnet" schools, this indicator would measure the ability of parents to conveniently get their children to the school location every morning. (U: It would be unthinkable to require parents to transport their children extremely long distances each morning to get them to school). Also, if children are expected to walk to school, school ought to be close by. (U: It would be unthinkable to expect children to walk a long distance to school). Since public schools are funded by tax money, which residents of the city must pay, it is important that citizens get what they pay for. (U: It is unthinkable that persons be charged tax money for inferior social services). Finally, education increases one's productivity and economic prospects. (N: Education increases one's economic chances in life). Spatial Obstacles? Yes. Particularly since this indicator deals with the safety of small children and the logistics of conducting them to school, any spatial barriers might bear strongly on the interpretability of this indicator. High-Intensity Uses? Yes. Junior high schools and middle schools are considered more intensive land uses than elementary schools, particularly in urban environments where troublesome and violent behaviors may be more likely. Indicator 19. Libraries Description: The distance from area centroids to libraries open to the public. Source: Switchboard telephone directory website. Calculation: Calculated via an ArcView Avenue macro language script which identifies the closest point feature in a geographic coverage to a point feature in another coverage. Value Should Be: Smaller. Justification: Libraries provide individuals free access to books, which is vital for educational purposes, and education increases productivity and economic prospects. This also saves individuals the money they would have spent buying these same books. (N: Libraries provide a basis for future productivity and economic prospects at no cost to those who use them). Spatial Obstacles? Yes. Spatial obstacles would have a strong effect on the use of libraries, which should be in an accessible location for optimum use.

76

High-Intensity Uses? Libraries represent perhaps the antithesis of the highintensity use. Indicator 20. Median Household Income Description: The median household income of an area. Source: 1990 United States Census. Calculation: None. This is the raw figure provided by the census. Value Should Be: Larger. Justification: This represents the amount of money earned in a household per year. Money can be used to directly purchase the necessities of life. (N: Money allows one to economically support oneself). Spatial Obstacles? No. High-Intensity Uses? No. Indicator 21. Median Property Value Description: The median value for property located in an area. Source: 1990 United States Census. Calculation: None. This is the raw figure provided by the census. Value Should Be: Larger. Justification: This measure has a special justification in that it is the direct representation of the market's valuation of the worth of property located in a certain area. Thus, insofar as the market can directly measure quality of life, this indicator fulfills this purpose. It is included in this database despite the fact that it fulfills neither of the two stated criteria for indicator selection because it is a traditionally recognized econometric measure of quality of life. Spatial Obstacles? No. High-Intensity Uses? No. Indicator 22. Nonviolent Crime Description: The number of crimes of a nonviolent nature taking place during the year 1994 per square mile in an area. Source: New Orleans Police Department. Calculation: Calculated by joining a geographic point coverage representing nonviolent crimes in the city of New Orleans with geographic polygon coverages representing city block group, tract and neighborhood areas. This yields a data set in which all of the points are coded by the area in which they are found. Then frequency statistics are calculated to get a number for each area corresponding to

77

the number of nonviolent crimes in each area. This number is then divided by the number of square miles in the area to normalize the figure to represent nonviolent crimes per square mile in the area. Value Should Be: Smaller. Justification: Crime erodes physical security, threatens health and safety, and causes property loss or damage. (N: Crime causes loss or damage of property; U: It would be unthinkable to promote greater levels of crime). Spatial Obstacles? No. High-Intensity Uses? No. Indicator 23. Parks (Centroid Distance) Description: The distance from an area centroid to the nearest centroid of a city park. Source: Gousha map of New Orleans; Rand-McNally New Orleans "street finder." Calculation: Calculated via an ArcView Avenue macro language script which identifies the closest point feature in a geographic coverage to a point feature in another coverage. This measure goes from the center of an area to the center of the park; it therefore measures the accessibility of the entire park, and not just part of the park. Value Should Be: Smaller. Justification: Parks have beneficial effects on human physical and mental health. (N: Parks are important for human health). Parks also provide a place for the enjoyment of natural scenery and recreation activities. (U: It would be unthinkable to have no appropriate place for "great outdoors" recreation activities). Spatial Obstacles? Yes. Spatial obstacles would have a strong effect on the use of parks. High-Intensity Uses? No. Though parks may be used by a large number of people for large-scale activities, they are generally looked upon as favorable additions to a neighborhood rather than nuisances. Indicator 24. Parks (Perimeter Distance) Description: The distance from an area centroid to the edge or perimeter of a city park. Source: Gousha map of New Orleans; Rand-McNally New Orleans "street finder."

78

Calculation: Iterative running of the "Select By Theme" ArcView menu choice, using the area centroids and park polygons as the two geographic themes, provided distance values for the indicator; these were subsequently entered into the database. This measure goes from the center of an area to the edge of a park; it therefore measures the accessibility of any part of a park rather than all of it. Value Should Be: Smaller. Justification: Parks have beneficial effects on human physical and mental health. (N: Parks are important for human health). Parks also provide a place for the enjoyment of natural scenery and recreation activities. (U: It would be unthinkable to have no appropriate place for "great outdoors" recreation activities). Spatial Obstacles? Yes. Spatial obstacles would have a strong effect on the use of parks. High-Intensity Uses? No. Though parks may be used by a large number of people for large-scale activities, they are generally looked upon as favorable additions to a neighborhood rather than nuisances. Indicator 25. Police Stations Description: The distance from an area centroid to the nearest police station or substation. Source: New Orleans Police Department; Metropolitan Crime Commission. Calculation: Calculated via an ArcView Avenue macro language script which identifies the closest point feature in a geographic coverage to a point feature in another coverage. Value Should Be: Smaller. Justification: Proximity to a police station would suggest the likelihood of prompt police protection in the case of a crisis situation. (N: Proximity to the police should increase the likelihood of public safety and the protection of human life). Spatial Obstacles? Yes. In addition, the city's police districts serve as limits for where each police station serves. High-Intensity Uses? Yes. Police stations are unquestionably high-intensity use facilities, with much associated noise and parking requirements. Indicator 26. Post-High-School Graduation Rate Description: The percentage of persons 18 or over graduating from a 2-year or 4year collegiate institution in an area. Source: 1990 United States Census.

79

Calculation: The number of people graduating from a 2-year or 4-year collegiate institution / the number of people 18 and over. Value Should Be: Larger. Justification: Higher levels of education makes more likely the possibility of high-paying employment. (N: Higher education leads to increased productivity and economic prospects). Spatial Obstacles? No. High-Intensity Uses? No. Indicator 27. Public Assistance Description: The percentage of persons in an area receiving public assistance. Source: 1990 United States Census Calculation: The number of people on public assistance / The total population of the area. Value Should Be: Smaller. Justification: The number of persons on public assistance, depending on one's view of it, is either a proxy measure for lack of economic opportunity (N: Public assistance measures the lack of economic opportunity in an area) or it is an indicator of the presence of a social pathology and a breakdown in moral values, for which the taxpayer must foot the bill (N: Public assistance payments require the raising of taxes). In either event, it is considered a success when public assistance levels do not increase. (U: It would be unthinkable to advocate that more people go on public assistance). Spatial Obstacles? No. High-Intensity Uses? No. Indicator 28. Real Employment Description: The percentage of employed members of the total population of an area. Source: 1990 United States Census. Calculation: The number of employed persons / The total population of an area. Value Should Be: Larger. Justification: Represents the economically dependent population of an area by the standards of the marketplace. This measure includes "discouraged workers," retired persons, disability pensioners, homemakers and children in its measure of employment, and thus gets at what percentage of the population actually is working at a paid job in the labor market to support those individuals who are outside it for whatever reason. As with employment, this is an indicator of the

80

economic health of a community. Those who are employed have the means to economically support themselves and dependents. (N: Employment is an economic necessity in the absence of inherited income or a benefactor). Spatial Obstacles? No. High-Intensity Uses? No. Indicator 29. Recreation Facilities Description: The distance from an area centroid to the nearest city recreation facility. Source: Switchboard telephone directory website. Calculation: Calculated via an ArcView Avenue macro language script which identifies the closest point feature in a geographic coverage to a point feature in another coverage. Value Should Be: Smaller. Justification: Recreation facilities have beneficial effects on human physical and mental health. (N: Recreation facilities are important for human health). Recreation facilities also provide a place for young people to get involved with intramural sports and to spend time without getting into trouble after school. (U: It would be unthinkable to have no appropriate place for young people to enjoy recreation activities; U: It would be unthinkable not to provide alternative activities for young people after school). Spatial Obstacles? Yes. Spatial obstacles would greatly affect the use of recreation facilities. High-Intensity Uses? Perhaps. It would depend on the nature of the facility. Stadiums, for example, would indeed invite a high intensity of use. Indicator 30. RTA To Work Description: Percentage of persons in an area taking either the bus or the streetcar (the two kinds of conveyance operated by the New Orleans "Regional Transit Authority," or RTA) to work. Source: 1990 United States Census. Calculation: (The number of person riding the bus + The number of persons riding the streetcar) / The total population of the area. Value Should Be: Larger. Justification: This particular indicator is justified so long as one takes it as a given that a city the size of New Orleans will have public transit. If this is given, then it is important that people ride the bus and streetcar so as not to waste the subsidy money that keeps the Regional Transit Authority afloat. (U: It would be

81

unthinkable to spend money on a bus and streetcar system few people ride). If more individuals did ride RTA, it would conserve fossil fuels as well. (N: Fossil fuels are necessary for economic production and public services and thus should be conserved). Spatial Obstacles? No. High-Intensity Uses? Possibly. Bus service in quiet residential neighborhoods may be pushed to a major arterial street. Indicator 31. Schools Description: The distance from an area centroid to the nearest school in the K-12 range. Source: Switchboard telephone directory website. Calculation: Calculated via an ArcView Avenue macro language script which identifies the closest point feature in a geographic coverage to a point feature in another coverage. Value Should Be: Smaller. Justification: This indicator measures the distance to the nearest school, whether public or private. It measures the extent to which areas are served by K-12 educators in any school system. Though this indicator does not measure the quality of the nearest school, it does measure whether some areas are not served by any sort of school whatever. (U: It would be unthinkable to have a neighborhood extremely far from any K-12 range school, public or private). Spatial Obstacles? Yes. Particularly since this indicator deals with the safety of small children and the logistics of conducting them to school, any spatial barriers might bear strongly on the interpretability of this indicator. High-Intensity Uses? Possibly, depending on the size and kind of the school in question. Indicator 32. Toxic Releases Description: Distance from area centroids to sites releasing toxic chemicals into the air, water or land. Source: Environmental Protection Agency Toxic Release Inventory Calculation: Calculated via an ArcView Avenue macro language script which identifies the closest point feature in a geographic coverage to a point feature in another coverage. Value Should Be: Larger.

82

Justification: Toxic chemicals discharged into air, water or land pose a health threat to the immediate environment of an area. (N: Toxic chemicals are potentially harmful to human health). Spatial Obstacles? No. High-Intensity Uses? No. Toxic sites are, of course, generally industrial plants with a tremendously high intensity of use. However, this indicator has a positive implication for quality of life the longer the distance from the site, so there is no concern about this being a "NIMBY-BANJO" indicator; indeed, toxic sites are better described as a wholly unwanted land use - a traditional "NIMBY" or "LULU." Indicator 33. Toxic Water Releases Description: Distance from area centroids to sites releasing toxic chemicals into the water. Source: Environmental Protection Agency Toxic Release Inventory Calculation: Calculated via an ArcView Avenue macro language script which identifies the closest point feature in a geographic coverage to a point feature in another coverage. Value Should Be: Larger. Justification: Toxic chemicals discharged into water pose a health threat to the immediate environment of an area. (N: Toxic chemicals are potentially harmful to human health). Spatial Obstacles? No. High-Intensity Uses? No. Toxic sites are, of course, generally industrial plants with a tremendously high intensity of use. However, this indicator has a positive implication for quality of life the longer the distance from the site, so there is no concern about this being a "NIMBY-BANJO" indicator; indeed, toxic sites are better described as a wholly unwanted land use - a traditional "NIMBY" or "LULU." Indicator 34. Traffic Index Description: A ratio of the total population of an area to a weighted array representing the number persons carpooling to work normalized by the measurement of the area in square miles. This is an index representing the amount of traffic congestion caused by trips to work begun in the area. Source: 1990 United States Census.

83

Calculation: (The total population of the area / (((Those driving alone) * 1) + ((Those in 2-person carpools)* 2) + ... + ((Those in 7-person carpools) * 7)))) / The number of square miles in the area. Value Should Be: Larger. Justification: Traffic is a nuisance. It causes a noise problem and lends to stress. (U: It would be unthinkable to promote more traffic in an area; N: Traffic noise and congestion leads to unhealthy levels of stress). High amounts of traffic makes areas less safe for pedestrians, small children and other drivers. (N: High amounts of traffic leads to the likelihood of physical danger being increased). Spatial Obstacles? No. High-Intensity Uses? No, though perhaps the NIMBY-BANJO relationship is with the major arterial road in this case - one wants to live near an arterial road, but not too near it. Indicator 35. Vacancies Description: Percentage vacant units of all housing units in an area. Source: 1990 United States Census. Calculation: The total vacant units in the area / The total housing units in the area. Value Should Be: Smaller. Justification: Vacant units often provide a base for criminal operations. They also lend to an image of area decline. (N: Vacancies provide opportunities for criminals to threaten public security; U: It would be unthinkable to promote vacancies in a neighborhood due to their association with neighborhood decline). Untended vacant units are also often fire hazards. (N: Vacant units can can catch on fire and in so doing threaten the lives and property of others). Spatial Obstacles? No. High-Intensity Uses? No. Indicator 36. Violent Crime Description: The number of crimes of a violent nature per square mile in an area. Source: New Orleans Police Department. Calculation: Calculated by joining a geographic point coverage representing violent crimes in the city of New Orleans with geographic polygon coverages representing city block group, tract and neighborhood areas. This yields a data set in which all of the points are coded by the area in which they are found. Then frequency statistics are calculated to get a number for each area corresponding to the number of violent crimes in each area. This number is then divided by the

84

number of square miles in the area to normalize the figure to represent violent crimes per square mile in the area. Value Should Be: Smaller. Justification: Crime erodes physical security, threatens health and safety, and causes property loss or damage. (N: Crime is life and health threatening; N: Crime causes loss or damage of property; U: It would be unthinkable to promote greater levels of crime). Spatial Obstacles? No. High-Intensity Uses? No. Indicator 37. Violent Nonproperty Crime Description: The number of crimes of a violent nature which are not propertyrelated crimes per square mile in an area. This measures the number of violent crimes which arise out of personal disputes rather than an economic motive. Source: New Orleans Police Department. Calculation: Calculated by joining a geographic point coverage representing violent nonproperty crimes in the city of New Orleans with geographic polygon coverages representing city block group, tract and neighborhood areas. This yields a data set in which all of the points are coded by the area in which they are found. Then frequency statistics are calculated to get a number for each area corresponding to the number of violent nonproperty crimes in each area. This number is then divided by the number of square miles in the area to normalize the figure to represent violent nonproperty crimes per square mile in the area. Value Should Be: Smaller. Justification: Crime erodes physical security, threatens health and safety, and causes property loss or damage. (N: Crime is life and health threatening; N: Crime causes loss or damage of property; U: It would be unthinkable to promote greater levels of crime). Spatial Obstacles? No. High-Intensity Uses? No. Indicator 38. Walking To Work Description: The percentage of persons who reported to the census that they regularly walk to their place of employment. Source: 1990 United States Census. Calculation: The number of persons walking to work / The total population of the area. Value Should Be: Larger.

85

Justification: This indicator measures a number of things which bear greatly on urban quality of life. It serves to measure whether an area is safe in which to walk. (U: It would be unthinkable to have unsafe streets). It also measures the possibility of reaching one's place of employment by pedestrian means, and thus serves as a proxy measure for proximity to one's place of employment. (U: It would be unthinkable to require people to live extremely far from where they work; N: It could cause economic hardship to require people to live extremely far from where they work, as suggested by the "spatial mismatch" literature). Finally, walking conserves fossil fuels and lowers automobile exhaust emissions. (N: Fossil fuels are necessary for economic production and public services and thus should be conserved; N: Automobile exhaust increases the likelihood of air pollution, which threatens human health). Spatial Obstacles? No. This is actually in part a measure of the lack of spatial obstacles, so such obstacles do not affect the measure's accuracy. High Intensity Uses? Yes. One may not wish to live right next to where one works, but it is important for economic reasons as well as reasons of convenience to live reasonably close to one's place of employment.

86

CHAPTER VI STATISTICAL METHODOLOGY

Four statistical techniques were used in the evaluation of the research hypotheses: Pearson product-moment correlations, principal component factor analysis, linear regression and curvilinear regression. This section of the thesis exposits the techniques and identifies the rationale behind their use in the research design.

Correlations The Pearson product-moment correlation (r) is a measure of the extent and direction of covariation of variables. Two variables are said to be correlated when a change in the value of one variable tends to be associated with a "consistent corresponding change" in the value of the other (Parsons, 1974). The formula for the calculation of r, where X and Y are the values of the two variables to be correlated, is presented below:

The square of r can be generally interpreted as a measure of the portion of the change in variance in one variable which is accounted for by covariance with another variable. Thus the r2 statistic is often referred to as a "coefficient of determination"

87

which shows the extent to which one variable "determines" the value of another. This term is misleading, as r2 does not provide any information on the direction of causality; indeed, a high r2 is not even sufficient to prove causality (King, 1986). All that a high r2 demonstrates is that a certain amount of covariance in a variable could be explained by reference to another variable; the latter variable may, however, be either causally unrelated to the former or the effect of the former rather than its cause. Correlations are only meaningful in the case of data which can be modeled as a linear relationship. A low value for r2 may not rule out the existence of covariance between variables. In many cases, scatterplots can suggest the possibility that certain relationships can be modeled as curvilinear relationships. Values for r need to be tested for significance by means of the Student's t test. This test measures the conditional probability that a relationship as strong as the one apparently observed in the data would be present, if the null hypothesis (in this case, that the data are not linear) were true. For sample data, all variables for which r is calculated must be drawn from a bivariate normal distribution. In such circumstances, r is considered an estimator of a population correlation coefficient r and thus an assumption that all variables are normally distributed must be posited. The project database represents the entire population of block groups, tract and neighborhoods in the city, however, so r is known to be equivalent to r in this case and no estimation for the population parameter is necessary.

88

In order to test H1, it is necessary to create a correlation matrix from the 38 indicator variables in the dataset and the study variable of home ownership. As stated in the hypothesis, values for the statistic r2, as calculated between the study variable and one of the indicator variables, which are over .3 will be considered evidence of a strong relationship between home ownership and the indicator. This will not establish the causal direction of the relationship, but since the hypothesis is that there is not a relationship to begin with, this will not matter in terms of verifying or falsifying the original hypothesis.

Factor Analysis Factor analysis utilizes matrix algebra to construct a number of orthogonal, or mutually uncorrelated, "factor" variables which serve to collapse and group a larger number of original variables. By so doing, factor analysis uncovers the independent "sources" of data variation (Rummel, 1970). The fundamental theorem of factor analysis (Thurstone, 1947) is expressed by the following equation:

In the above equation, the matrix R is the correlation matrix of all variables upon which factor analysis is being performed. It is a symmetric matrix, since each of the variables is being correlated with itself and all the other variables. The matrix I is the "identity matrix," or a matrix entirely composed of zeroes except along the principal

89

diagonal, where the values are all ones. The matrix H2 is a matrix of "communalities," or proportions of the variance of each variable held in common with all of the other variables in the dataset. The value of the communalities are found in the principal diagonal of this matrix. The matrix F represents the "factor loadings" for each of the m variables on each of the p factors produced in the process of factor analysis. Factor loadings represent how much each variable contributes to each of the factors, and are analogous to standardized scores, or "z-scores," in a normal curve distribution. The F' matrix is the matrix transpose, or reversal of rows and columns, of the F matrix. The fundamental theorem of factor analysis is a circular theorem, as one needs to be able to calculate what portion of the variance of variables in the data matrix are held in common before one may calculate the factor loadings, but one needs to know the value of the factor loadings before one may calculate which portions of variance in the data matrix are held in common. In order to get around this fault of the model, an estimation technique called component factor analysis was fashioned. Component factor analysis simplifies the theorem to the following equation:

When the communalities of variables with other variables in the matrix approach 1, the correlation matrix R is equivalent to the matrix equation R-I+H2, and thus R may replace R-I+H2 in the factor analysis theorem. A method of component factor analysis, called principal components analysis, provides initial estimation of the value of the

90

communalities as 1 in order to begin an iterative process called "convergence.” In this process, factor loadings are calculated and then used to recalculate the communalities until both the loadings and the communalities reach stable values. Principal components factor analysis proceeds from the assumption that the more data clusters in one location, the more valid the argument that one may proceed from the simplified theorem. In order to determine the extent to which separate "factors" can be identified as clusters of the original data, the principal axes technique is often used. This technique, first used in the analysis of psychology data (Hotelling, 1933), identifies the minimum number of orthogonal dimensions required to linearly reproduce the original data (Rummel, 1970). This involves the fitting of a hyperellipsoid figure around the extent of a data cluster, which is measured by a scalar value called an eigenvalue. The greater the extent of the data cluster, the greater the eigenvalue, which is equivalent to the length of the major principal axis of the hyperellipsoid. It is also analogous to a regression line for the data cluster. A two-dimensional graphic exposition of this concept is presented as Figure 4. To ensure the greatest possible orthogonality, or mutual lack of correlation, of each of the identified factors, a vector called the eigenvector is identified for each calculated factor. The eigenvector may be represented as a minor principal axis of the hyperellipsoid at right angles to the major principal axis representing a calculated factor.

91

Since it is at right angles to this axis, it allows for the calculation of a factor which does not share characteristics with that factor. A graphic representation of the important geometric concepts behind principal components analysis is provided (Figure 4) for a greater clarification of the process, which may continue until the factors either account for a trivial amount of variance or are attributable to random error. For purposes of interpretability, it is often desirable to use some form of rotation of the variable axes to make factor correlations with the original variables and variable loadings on each factor more substantively meaningful (Norušis, 1990; Rummel, 1970). The "varimax" criterion for variable axis rotation is considered the best function for producing factors with a clear substantive interpretation. The varimax rotation procedure rotates variable axis according to an algorithm which minimizes the number of variables which have high loadings on a factor. This results in the calculation of factors only a few of which have large values for correlations with certain variables or high loadings for certain variables. When high correlations or loadings occur, then, the implication of this is clear to the researcher. A variable which has a high loading on a factor after varimax rotation is a variable which greatly contributes to that factor. Factors resulting from analyses rotated according to the varimax criterion are, thus, more readily interpretable as

92

Figure 4. Geometric Concepts in Principal Components or Principal Axes Factor Analysis Source: Rummel, Applied Factor Analysis.

93

being functions of certain variables more than others. In order to test H2, it will be necessary to run a principal components factor analysis on the 38 indicator measures utilizing varimax rotation so that the factors produced by the analysis will be more readily interpretable as functions of certain variables in the indicator database. The factors which come out of the analysis, since they represent functions of certain variables, will therefore provide a rationale for the grouping of the indicators according to their common fluctuations over all areas in each of the datasets.

Linear Regression In order to evaluate H3, the information in the correlation matrices for each level of spatial aggregation will generally suffice. However, regression analysis is necessary to further elaborate the relationship between home ownership and the indicators at each level of geographic scale. Hypothesis H3, though specifically concerned with verifying a presumption of the urban sustainability literature, is more generally concerned with modeling the differences in the benefit home ownership confers on indicator scores at the three spatial levels of aggregation. Linear regression analysis can be used to compare the relationship of one or more independent, explanatory variables to the same dependent variable (King, 1986). It can also be used to test for the linearity of a relationship through the analysis of variance

94

procedure, also referred to as ANOVA, which compares distance of points to the regression line to the distance of points from the mean of the dependent variable (Parsons, 1974). The evaluation of hypothesis H3 requires the comparison of relationships between home ownership and indices or factors based on the 38 quality of life indicators across the three levels of spatial aggregation. The linearity of these relationships will also have to be established, and if it cannot be established at one of the levels, a curvilinear regression procedure may be more appropriate to modeling the data. Also of importance is the consideration of variables which are collinear with home ownership and which may also explain a variation in indices of quality of life. In particular, the median household income variable, which also happens to be one of the indicators, needs to be tested for its relationship to the other variables in the indicator database. The regressions done in this thesis project, when they involve the median household income variable, eliminate any indicator in which median household income plays a role in its calculation, so that the regressions measure only median household income's relation to indices or factors representing the remaining indicators in the database. Multiple regression can provide information on both the relationship of home ownership and of median household income on changes in index values or factor

95

loadings. The form taken by a multiple regression line function with one dependent variable X1 and two independent variables X2 and X3 is the following:

The term on the left side of this equation represents the conditional mean of X1 associated with a specified value of X2 and X3. The first term on the right side represents the conditional mean of X1 if X2 equals 0 and X3 equals 0. This is the constant term of the regression equation. The next terms on the right hand side represents the amount (signified by the b coefficients in these terms of the equation) that the conditional mean of X1 must change as proportions of X2 or X3 for every unit change in X2 or X3. These are called the regression coefficients (Parsons, 1974). Regression coefficients can be turned into standardized scores so that they may be made comparable in terms of their sizes (Younger, 1979: 369). The scores are called standardized regression coefficients, or “beta weights,” and measure the direction and number of standard deviations the dependent variable changes for each one standard deviation increase in [the independent variable (Younger, 1979: 371). This is useful when comparing variables which can be represented in terms of the same units of measure, like miles or pounds. It frequently misrepresents data, however, which cannot be measured in the same units. A mile, for example, cannot be compared to a pound, whether miles and pounds are converted to standard deviations of the variable distributions or not (King, 1986: 669-672.) Standardized regression coefficients are often

96

used to interpret regression data based on analysis of scales or indices which have arbitrarily set values, such as those which are often employed in social science (Younger, 1979: 371). These coefficients are represented as b when they refer to a sample group, and in this case the value of the coefficient would be an estimate of the population parameter b. Once again, since the project database represents the entire population of block groups, tract and neighborhoods in the city, b is known to be equivalent to b and therefore no estimation for the population parameter is necessary. Any variable subscript in the preceding equation to the right of the period represents a variable held constant in the analysis. To put this another way, the term on the left side of the equation represents the conditional mean of X1 holding X2 and X3 constant. An analysis of variance performed on a multiple regression equation allows one to test whether the null hypothesis that the value of R2, the multiple regression coefficient, is equal to 0 and that there is therefore no linear relationship. Tests of significance can also be applied to test whether any of the b coefficients are equal to 0 in order to verify their suitability as elements of the regression equation. In the event that a relationship is not linear, curvilinear regression might provide a better model for the data.

97

Curvilinear Regression Fitting a curve to data representing home ownership's relationship to quality of life indices would be best accomplished by a polynomial regression model. The SPSS software package provides algorithms for three kinds of polynomial regression models, the linear model (the standard regression model, which graphs as a line), the quadratic model (which graphs as a parabola) and the cubic model (which graphs as an inverse curve). The latter of these three models was chosen as a method for analysis of curvilinear relationships between quality of life indicators and home ownership. The equation for the cubic polynomial model of regression is provided below:

In the above equation, X represents the one independent variable which is taken to the first, second and third exponential power and Y represents the dependent variable. A cubic polynomial is a polynomial expression of the third order. The graphical representation of a third-order polynomial can have as many as two “turning points” where the direction of the curve changes. This is the case because any nth-order polynomial can only have as many as n-1 turning points (Lial, Hornsby and Schneider, 1997). The curve need not have turning points, but may have only as many as two according to the constraints of this model. What this means is that the cubic model can represent up to two distinct changes in the direction of a relationship between variables, from negative to positive or vice-

98

versa. Regarding the variables with which this study concerns itself, quality of life indicators may correlate strongly and positively, for example, with a certain range of levels of home ownership, but the relationship may drop off at higher levels of home ownership instead of continuing to climb. The cubic model can establish the differences in the relationship at specific levels of home ownership and provide a rationale for further linear regressions on subsets of the data within certain ranges of home ownership.

99

CHAPTER VII RESULTS

The results of statistical analysis verify that home ownership, if it plays a role in neighborhood quality of life, plays a small role at best. As expected, home ownership does not seem to strongly and negatively affect an area, but the positive benefit to neighborhood quality of life which many of home ownership’s defenders presume is not of a magnitude to justify many of their claims. The results of this analysis provide a strong rationale for de-emphasizing home ownership programs as the centerpieces of policy they have become, and for offering in their place programs which offer home ownership assistance as only one of many options for neighborhood revitalization. The results from the factor analysis confirm that, while home ownership does seem to have a strong positive relationship with indicators and factors that are primarily economic in nature, they do not positively affect indicators that are non-economic. This suggests that home ownership strategies favor the economist’s view of what constitutes neighborhood revitalization and ignore or even aggravate concerns which are noneconomic but nonetheless important and justifiable to neighborhood residents. This also provides a rationale for policy which is more catholic and inclusive of other revitalization initiatives besides home ownership strategies. Factor analysis suggests the “three E’s” model provided by the urban sustainability paradigm is not completely supported, though arguably two of the “E’s”

100

were identifiable in the results. An “economic” factor was clearly identifiable. The “environmental” factor hoped for was present to an extent in the results, though alternative interpretations of this factor are possible. Two other factors might together approximate the predicted “social equity” factor, though important differences between the factors suggest that this is an overly simplistic inference to draw from the data. Two factors were important in each of the factor analyses, the “economic” and “distance to services” factors. Though this finding did not mirror the expectations of the sustainability paradigm, it was not far removed from the main thrust of its theory. Home ownership has a fairly strong positive relationship with the “economic” factor, but a moderate negative relationship with the factor representing “distance to services,” rather than the decisively strong and positive relationship advocates of home ownership might claim. The results of correlation analysis run against the expectation of the hypothesis concerning the role of geographic scale in home ownership’s benefit regarding the indicators. Contrary to the original hypothetical supposition, no group of indicators becomes “more important” at a different level of geographic aggregation. However, at higher levels of aggregation, the benefits of home ownership in terms of the indicators do seem more apparent. This suggests that home ownership does have important residual effects in outlying areas; such a conclusion is consistent with those made in the literature (Galster, 1987; Residential Mobility and Public Policy, 1980).

101

The following sections of the chapter set forth the bases for these general findings.

Home Ownership and The Indicators The following table (Table 7) represents a correlation matrix of all of the indicator variables with the variable representing the percentage of home ownership for all of the census block groups, census tracts and city neighborhoods in the city of New Orleans. The table represents the values for r in a correlation matrix of the individual indicator variables with the percentage home ownership variable. The original hypothesis H1 specified that no value for r2 would be over .3. Taking the square root of r2, this would mean that the preceding table can help to verify the hypothesis by setting the critical values for r at over +.548 or under -.548. At the block group level, home ownership was only correlated with one variable at the level sought by H1, the Median Household Income indicator. It was expected from the outset that this particular indicator would correlate very highly with home ownership. In Figure 5, the similarity of the spatial distribution of the Median Household Income indicator is shown. A number of variables fell just short of the prescribed levels for r of +.548 or -.548; of these, many of these variables are either themselves indicators of economic affluence or are indicators which are closely covariant with income. Vacancies

102

Table 7. Values for r in Quality of Life Indicators Correlation Matrix with Percentage Home Ownership Indicator Variable Should Be Block Group Tract Neighborhood Value Value Value Bicycling To W ork Larger -.080* .009 -.157 CAT Elem entary Test 50th Percentile+ Sm aller -.022 .125 .046 Clothing Stores Sm aller .217** .285** .172 Com m uting Index Sm aller .105** .257** .018 Diversity of Em ploym ent Sm aller -.048 .018 -.156 Doctors (Centroid) Sm aller .199** .266** .130 Doctors (Point-in-Polygon) Larger -.089* -.069 -.116 Em ploym ent Larger .406** .505** .552** Fire Stations Sm aller .224** .315** .287* Graduate Exit Exam 90%+ Sm aller .233** .209** .122 Grocery Stores (Centroid) Sm aller .291** .294** .353** Grocery Stores (Point-in-Polygon) Larger -.296** -.340** -.498** Health Risk Larger -.286** -.184* -.235* High School Graduation Rate Larger .495** .528** .551** Hom e Affordability Sm aller -.167** -.111 -.377** Hospitals Sm aller .470** .451** .418** LEAP Elem entary Test 90%+ Sm aller .030 .146* .018 LEAP Jr./Middle Test 90%+ Sm aller .088* .123 .019 Libraries Sm aller .214** .251** .154 Median Household Incom e Larger .625** .682** .735** Median Property Values Larger .355** .440** .418** Nonviolent Crim e Sm aller -.274** -.302** -.454** Parks (Centroid) Sm aller .093* .205** .014 Parks (Perim eter) Sm aller .078* .194** .034 Police Stations Sm aller .264** .331** .181 Post-High School Graduation Rate Larger .353** .410** .348** Public Assistance Sm aller -.393 -.399** -.614** Real Em ploym ent Larger .297** .368** .430** Recreation Facilities Sm aller .353** .346** .277* RTA To W ork Larger -.286** -.168* -.478** Schools Sm aller .146** .234** .145 Toxic Releases Larger -.287** -.201** -.230 Toxic W ater Releases Larger -.280** -.191** -.219 Traffic Index Larger -.339** -.321** -.158 Vacancies Sm aller -.507** -.349** -.642** Violent Crim e Sm aller -.430** -.523** -.642** Violent Non-Property Crim e Sm aller -.438** -.520** -.638** W alking To W ork Larger -.252** -.241** -.285* * Significant at the .05 level. ** Significant at the .01 level. Cells show correlations for r; correlations meeting condition of r 2 > 3 are shown in bold.

103

Figure 5. Median Household Income in New Orleans by Census Block Group (quintile distribution)

104

(r = -.507, r2 = .257), High School Graduation Rate (r = .495, r2 = .245), Violent NonProperty Crime (r = -.438, r2 = .192), Violent Crime (r = -.430, r2 = .185) and Employment (r = .406, r2 = .165) are all affected fairly strongly and in a favorable direction regarding quality of life. As can be seen in Table 8, of these variables High School Graduation Rate correlates very strongly (r = .663, r2 = .440) and the others correlate moderately strongly with Median Household Income (Vacancies: r = -.303, r2 = .092; Violent Non-Property Crime: r = -.331, r2 = .110; Violent Crime: r = -.263, r2 = .069; Employment: r = .454, r2 = .206). One interesting indicator in the correlation matrix at the block group level was the Hospitals indicator. Home ownership influenced this variable in an unfavorable direction, and although it did not do this to the extent required by H1, it did do so fairly strongly (r = .470, r2 = .221). This means that a high percentage of home ownership in a block group increases the distance to the nearest hospital by a fairly large amount. Far from increasing convenience, increased home ownership impedes people from being able to access a critically and necessarily proximate health resource. The map presented as Figure 6 demonstrates this disturbing relationship spatially. One can conclude that even the issues surrounding high-intensity land use and the NIMBY-BANJO phenomenon are not sufficient to explain why home owners would want to live as far as they clearly do from the nearest emergency room. Though the

105

Table 8. Values for r in Quality of Life Indicators Correlation Matrix with Median Household Income Indicator Variable Should Be Block Group Tract Neighborhood Value Value Value Bicycling To W ork Positive -.028 .113 .019 CAT Elem entary Test 50th Percentile+ Negative -.159** -.116 -.077 Clothing Stores Negative .095* .078 .104 Com m uting Index Negative -.080* .012 -.241* Diversity of Em ploym ent Negative .145** .096 -.103 Doctors (Centroid) Negative .043 .074 .044 Doctors (Point-in-Polygon) Positive -.028 .000 -.004 Em ploym ent Positive .454** .528** .558** Fire Stations Negative .154** .199** .232 Graduate Exit Exam 90%+ Negative .083* .055 .038 Grocery Stores (Centroid) Negative .258** .159* .281* Grocery Stores (Point-in-Polygon) Positive -.258** -.316** -.427** Health Risk Positive .033 .063 .032 High School Graduation Rate Positive .663** .675** .721** Hom e Affordability Negative -.171** -.086 -.248* Hospitals Negative .205** .166* .243* LEAP Elem entary Test 90%+ Negative .057 .054 .036 LEAP Jr./Middle Test 90%+ Negative .155** .098 .051 Libraries Negative .063 .065 .052 Median Property Values Positive .705** .770** .803** Nonviolent Crim e Negative -.080* -.044 -.204 Parks (Centroid) Negative -.007 .057 -.071 Parks (Perim eter) Negative -.087* -.018 -.024 Police Stations Negative .223** .164* .146 Post-High School Graduation Rate Positive .680** .775** .698** Public Assistance Negative -.471** -.516** -.641** Real Em ploym ent Positive .476** .482** .481** Recreation Facilities Negative .226** .206** .212 RTA To W ork Positive -.281** -.282** -.509** Schools Negative .083* .098 .148 Toxic Releases Positive .032 .061 .023 Toxic W ater Releases Positive .034 .054 .047 Traffic Index Positive -.294** -.256 -.143 Vacancies Negative -.303** -.258** -.462** Violent Crim e Negative -.263** -.369** -.478** Violent Non-Property Crim e Negative -.331** -.440** -.518** W alking To W ork Positive .012 .018 .037 * Significant at the .05 level. ** Significant at the .01 level. Cells show correlations for r; correlations m eeting condition of r 2 > 3 are shown in bold.

106

Figure 6. Block Group Centroid Proximity to Hospitals (quintile distribution)

107

correlation did not satisfy the specifications of H1, it is high enough in the wrong direction to be very disquieting. The lack of nearby hospitals is, furthermore, clearly more associated with increased home ownership than of increased income. It is not the rich person who is far from a hospital, but rather the home owner. At higher levels of geographic scale, contrary to the assumptions of H3, there is no change whereby home ownership begins to correlate with a different group of indicators. At the tract and neighborhood levels, home ownership correlates for the most part with the same group of indicators but more strongly. At the tract level, once again, only Median Household Income correlates with home ownership highly enough (r = .682, r2 = .465) to satisfy H1. The list of variables moderately correlating with home ownership is the same, with Median Property Values (r = .440, r2 = .194) and Post-High School Graduation Rate (r = .410, r2 = .168) coming into greater prominence. At the neighborhood level, more variables meet the test of H1 and have a value for r2 which is over .3. Indicators which meet the test are Employment (r = .552, r2 = .305), High School Graduation Rate (r = .551, r2 = .304), Median Household Income (r = .735, r2 = .540), Public Assistance (r = -.614, r2 = .377), Vacancies (r = -.642, r2 = .412), Violent Crime (r = -.642, r2 = .412) and Violent Non-Property Crime (r = -.638, r2 = .407). Each of these indicators also correlates very highly with Median Household Income; the lowest correlation of any of these indicators with Median Household Income is that of the Vacancies indicator, which has a value for r of -.462, or an r2 of .213. The neighborhood data therefore well confirms that the indicators most associated with high

108

levels of home ownership are either themselves “economic indicators” or indicators which are strongly associated with income level. An indicator which becomes important at the neighborhood level, though it does not correlate at a level which meets the requirements of H1, is the RTA To Work indicator (r = -.478, r2 = .228). This indicator associates an unfavorable outcome for quality of life with home ownership; the higher the level of home ownership, the less people ride city buses or streetcars to work. This relationship is illustrated on the following page by Figure 7. However, the intensification of this relationship at the neighborhood level is consistent with a similar intensification of the relationship between the RTA To Work indicator and Median Household Income at the neighborhood level (r = -.509, r2 = .259). Thus it is likely the case that income is the important explanatory variable here as well; buses are too declassé for the well-to-do. Home ownership probably does not explain a lack of bus service, but it also apparently does little to facilitate it. In general, one can glean from this data that home ownership does not have a profound benefit in terms of all manner of quality of life indicators. Where there is a fairly high degree of association with certain indicators, these indicators are either narrowly economic in their focus or themselves strongly associated with income level. The expectation of H1 that no indicators would be strongly associated with home

109

Figure 7. Percentage Riding RTA in New Orleans by Census Block Group (quintile distribution)

110

ownership was confirmed by the results of statistical analysis, in that only the Median Household Income indicator correlated with home ownership strongly at the block group and tract levels. At the neighborhood level, H1 fared less well, as 7 of the 38 indicator variables were strongly associated with home ownership by the requirements of the hypothesis. For the remaining variables, even at the neighborhood level, the hypothesis was confirmed. The relationships between home ownership and the indicators generally intensified at larger geographic scales; this ran contrary to the assumption of H3 that certain indicators would come into play at specific levels of aggregation and not play any role at other levels of aggregation. Indeed, only one group of indicators plays a role at any level of aggregation, and it is evidently the same group. It is clear from the above data that H3 has not been confirmed. In the following section the contribution of geographic scale to relationships between home ownership and groups of indicators will be elaborated, but one can see at this point that this hypothesis clearly fails.

Factoring of the Indicators The supposition of H2 was that, following the “three E’s” model of the urban sustainability paradigm, three groups of indicators would emerge from factor analysis and correspond to the “economic,” “social equity” and “environmental” groups in that model.

111

The factor analyses done for this thesis did partially confirm the model, at least with respect to a clear “economic” group. However, the other two hypothesized groups could not be identified as clearly. The hypothesized “social equity” group failed to materialize, though a few of the factors which did emerge from the analysis, taken together, covered the indicators this group was expected to cover. The “environmental” group, on the otherhand, did seem to materialize in one of the factors, though alternative interpretations of the factor are possible. Table 9 shows the correlation matrix generated by factor analysis at the level of the census block group. The analysis generated 10 factors, which collectively explain 75% of the variance. The matrix is organized according to which indicator measures correlated with the factors at a level where r2 is over .3. Factor 1 correlates highly with 12 of the original indicators, and explains by itself 24.4% of the indicator variance. It is best described as a “distance to services” factor; each of the indicators correlating highly with this factor is a centroid distance measure from block group centroids to important services. The direction of the relationship is unfavorable with regard to quality of life; the distances represented by this factor are long rather than short. This factor represents a “spatial mismatch,” whereby people live far away from services they need. One section of the city of New Orleans, located in the northwest corner of the

112

Table 9. Block Group Level Correlation Matrix from Factor Analysis
Indicators Schools D octors (C entroid) G roceries (C entroid) Libraries C lothing Police Stations R ec Facilities LEAP Elementary H ospitals Fire Stations G rad Exam C AT Elementary H igh School Post H igh School Median Income R eal Employment Employment Median Value Public Assistance H ealth Risk T oxic W ater R eleases T oxic R eleases Violent C rime Violent/N onproperty N onviolent C rime G roceries (Points) R T A T o W ork C ommute W ave Parks (C entroid) Parks (Perimeter) Bike T o W ork Employment D iversity W alk T o W ork H ome Affordability D octors (Points) LEAP Junior/Middle Vacancies T raffic Index O ne 0.872 0.847 0.841 0.836 0.819 0.769 0.739 0.708 0.696 0.621 0.559 0.556 0.081 0.021 0.101 0.008 -0.01 0.022 -0.07 -0.14 -0.15 -0.19 -0.14 -0.13 -0.12 -0.10 -0.15 0.048 0.291 0.328 -0.03 -0.05 -0.04 -0.07 -0.04 0.52 -0.05 -0.20 T wo -0.013 -0.062 0.106 -0.048 -0.015 0.153 0.151 0.026 0.116 0.088 0.049 -0.183 0.882 0.819 0.814 0.781 0.758 0.699 -0.562 0.065 0.069 0.067 -0.121 -0.245 0.086 -0.135 -0.12 0.039 -0.023 -0.071 0.033 0.362 0.134 0.006 0.007 0.096 -0.239 -0.17 T hree -0.042 -0.137 -0.011 -0.24 -0.138 -0.183 -0.183 -0.017 -0.18 -0.015 -0.488 -0.172 0.005 0.186 0.020 -0.079 -0.042 0.201 -0.016 0.967 0.963 0.958 0.033 0.032 0.081 0.068 0.130 -0.349 -0.139 -0.176 0.020 0.059 0.004 0.130 0.023 -0.225 0.053 0.271 Four -0.045 -0.034 -0.065 -0.08 -0.05 -0.158 0.074 0.027 -0.134 -0.202 -0.095 0.104 -0.077 -0.052 -0.099 -0.153 -0.087 -0.006 0.298 0.074 0.072 0.069 0.894 0.867 0.618 0.594 0.094 0.157 -0.061 -0.024 0.040 0.071 0.205 0.158 -0.03 0.004 0.355 0.427 Five -0.02 -0.006 -0.158 -0.045 -0.052 -0.073 -0.16 0.140 -0.123 -0.027 -0.044 0.302 -0.054 -0.265 -0.254 0.293 0.434 -0.19 0.258 0.031 0.030 0.025 0.056 0.082 -0.029 0.209 0.732 0.572 -0.028 0.089 -0.073 0.126 -0.045 0.201 -0.055 -0.128 0.431 0.493 Six 0.163 0.006 0.018 0.075 0.209 0.045 0.140 0.184 0.001 -0.004 0.255 0.137 -0.043 -0.037 -0.058 0.004 0.014 -0.089 -0.113 -0.087 -0.088 -0.097 -0.023 0.013 -0.002 -0.068 0.021 0.104 0.870 0.853 -0.039 -0.064 -0.082 -0.094 0.172 0.065 -0.076 -0.035 Seven 0.013 0.034 -0.019 -0.009 0.011 -0.041 -0.08 -0.012 -0.075 -0.041 -0.04 0.052 0.091 0.220 -0.13 0.213 0.075 -0.131 -0.051 0.024 0.024 0.021 0.057 0.019 0.051 0.021 0.035 -0.027 -0.058 -0.044 0.911 0.743 0.001 -0.075 0.105 0.000 -0.027 -0.066 Eight 0.054 0.014 -0.002 -0.013 -0.079 -0.081 -0.075 0.033 -0.128 -0.007 -0.062 0.047 0.071 -0.02 -0.127 0.281 0.109 -0.202 -0.153 0.010 0.014 0.007 0.194 0.056 0.443 -0.177 -0.003 -0.358 -0.05 -0.058 -0.13 0.263 0.764 -0.168 0.296 -0.015 0.269 -0.014 N ine -0.01 -0.074 -0.063 -0.032 0.015 -0.019 -0.009 0.013 -0.095 -0.01 -0.047 0.111 0.009 0.043 -0.068 -0.045 0.012 0.428 0.128 0.058 0.057 0.057 -0.007 -0.085 0.131 0.141 0.074 -0.051 0.032 0.030 0.018 -0.01 0.018 0.815 0.674 0.003 0.147 -0.001 T en 0.069 0.068 -0.024 0.056 0.143 0.071 -0.191 0.430 -0.466 0.023 0.425 -0.073 -0.042 0.057 0.029 0.027 -0.079 0.101 0.031 -0.024 -0.038 -0.028 -0.028 -0.042 -0.029 0.089 -0.039 -0.097 0.073 0.007 0.003 0.000 0.015 0.095 -0.08 0.688 0.278 0.109

(Cells show correlation values for r; boxes show correlations where r 2 > .3).

113

city, is perhaps the best example of the separation from services this factor models. This area, called Lakeview, is at once the most affluent neighborhood area in the city and the most secluded from the daily commerce of human activity. It is, perhaps, the best example of a quasi-suburban “bedroom community” in the city. Figure 8 demonstrates the lack of connection the Lakeview area has to a number of the locations used to calculate the quality of life indicators. On the map, the Lakeview area is in the northwest corner along Lake Pontchartrain and the Jefferson Parish border. Factor 2 correlates highly with 7 of the original indicators, and is the “economic” indicator predicted by H2. All of the indicators which the factor covers are all clearly economic measures except for two, the High School Graduation Rate and Post-High School Graduation Rate indicators. These two educational attainment indicators are highly covariant with Median Household Income. Factor 3 could arguably be the hypothesized “environmental” indicator. The three indicators calculated from the Environmental Protection Agency’s Toxic Release Inventory data correlate very highly with the factor, and in a direction favorable to quality of life. The postive value of the Toxic Releases, Toxic Water Releases and Health Risk measures indicate that toxic release sites are far from block groups which load highly on Factor 3. However, one can not be too confident of this result, as toxic release sites in the city of New Orleans are clustered primarily in New Orleans East, on the extreme eastern fringe of the city. This being the case, one could as well make the case that the factor

114

Figure 8. The Isolation of Lakeview

115

represents “Western-ness” as much as environmental safety. In order to make a case for the factor being a true “environmental” indicator, it is necessary to rely on some of the corroborating information provided by the correlation matrix. The Traffic Index indicator provides some limited support for the case; it is a moderately strong relationship in a direction favorable to quality of life. This means traffic congestion is low, and emissions from automobiles should therefore affect areas loading highly on Factor 3 less. The Commuting Time indicator is also favorably associated with the factor. The correlations for distance measures are all favorable, which indicates that areas loading highly on Factor 3 are close to important services. Although some of these services do not help establish Factor 3 as an “environmental” factor, some in particular, such as those relating to schools and libraries, do indicate that areas locating highly on Factor 3 might be fairly pleasant environments. The above defense does not clearly establish the factor as the sought-after “environmental” factor beyond all doubt. It would be interesting to see if studies of other cities, particularly those cities where environmental hazards such as toxic release sites are less clustered, generate more conclusive “environmental” indicator factors. Factor 4 is clearly a “crime” factor, with all three of the crime indicators correlating highly with the factor. Also correlating highly with the factor is the point-inpolygon measure for grocery stores, which measures the number of grocery stores in an area. The areas of the city which have the highest amount of crime are older areas of the

116

Figure 9. Commuting Waves in New Orleans by Census Block Group

117

(quintile distribution) city which also have a number of traditional “corner grocery stores” and convenience marts. Traffic is also evidently fairly low in areas loading highly on Factor 4, though this may be due to fewer people owning cars. The Public Assistance indicator shows a moderately large association with Factor 4, indicating higher numbers of people on public assistance. An association of a moderately high number of vacancies with the factor is also indicated. Factor 5 might best be reckoned a “working class commuters” factor. Commuting and riding RTA are very highly associated with this factor. It is clear that the factor represents a largely employed (according to the Employment indicator) group of people who don’t make a large amount of money (according to the Median Household Income indicator) and who either carpool or take the bus to work (according to the Traffic Index indicator). A map of the Commute Wave indicator is provided as Figure 9 to provide a point of reference; the area in darkest red closest to the Mississippi River, in the Lower Ninth Ward neighborhood, is the area which Factor 5 most closely approximates in its characteristics. The first five factors explain 58.8% of the variance by themselves, and the remaining five represent only 16.2%. Factor 6 is associated with extreme distance from parks. Factor 7 describes areas with high bicycle usage rates and a low diversity of employment, and probably best describes university areas. Factor 8 describes areas in

118

“walking distance” from work, such as the French Quarter and Central Business District areas of the city, where residences and places of employment are within close proximity and design practices accommodate heavy pedestrian traffic. Factor 9 is noteworthy for describing areas of the city where the value of housing is vastly outpacing the income of residents; the affordability of homes in areas loading highly on this factor is very low. On the other hand, these areas apparently are popular locations for doctor’s offices. Factor 10 is best interpreted as a “far from good schools” factor. Though not particularly far from schools in general, areas which load highly on this factor are notably far from any school which performed adequately on the LEAP, CAT or Graduate Exit standardized tests. The tract level data generated 8 factors in factor analysis, which were generally similar to the block group level factors. The correlation matrix for the tract level factor analysis is presented as Table 10. These factors account for 79.2% of the indicator variance, and 56.8% of the variance is explained by the first three factors. Factors 1 through 3 were more or less identical in interpretation to the factors from the block group level analysis. There were some changes in the remaining five factors, however, which are notable. Factor 4, which was a “crime and blight” factor at the block group level, changed slightly at the tract level, with the Nonviolent Crime indicator correlating somewhat less highly with the factor (r = .408, r2 = .166) and the Employment Diversity indicator coming more into play (r = .727, r2 = .529). Factor 4 at the tract level

119

Table 10. Tract Level Correlation Matrix from Factor Analysis
Indicators Schools C lothing D octors (C entroid) Libraries G roceries (C entroid) LEAP Elementary Police Stations R ec Facilities G rad Exam C AT Elementary H ospitals LEAP Junior/Middle Fire Stations Parks (Perimeter) Parks (C entroid) Post H igh School Median Value Median Income H igh School H ealth Risk T oxic W ater R eleases T oxic R eleases Violent C rime Violent/N onproperty T raffic Index Employment D iversity H ome Affordability Vacancies Public Assistance Employment R eal Employment W alk T o W ork D octors (Points) G roceries (Points) C ommute W ave R T A T o W ork N onviolent C rime Bike T o W ork O ne 0.9537 0.938 0.9351 0.9274 0.9138 0.8957 0.8823 0.8726 0.7972 0.7951 0.7829 0.7761 0.7478 0.7078 0.667 0.0307 -0.0032 0.0695 0.0903 -0.1468 -0.1679 -0.2481 -0.1768 -0.1597 -0.1644 0.0014 -0.0786 0.1093 -0.086 0.0082 0.02 -0.0289 -0.0505 -0.165 0.1662 -0.238 -0.1729 -0.1106 T wo -0.0013 -0.0085 -0.003 -0.0371 0.0512 -0.0129 0.1082 0.119 0.0171 -0.2314 0.0361 0.1177 0.093 -0.0747 0.0092 0.905 0.8696 0.8442 0.7216 0.1283 0.113 0.1328 -0.1361 -0.283 -0.1072 0.2992 0.1689 -0.1137 -0.5286 0.4868 0.4568 0.027 0.0282 -0.168 -0.0351 -0.3376 0.1318 0.4004 T hree 0.0283 -0.1106 -0.0197 -0.1701 0.009 -0.0912 0.119 -0.0843 -0.4123 -0.0397 -0.125 -0.2057 -0.019 -0.2076 -0.2024 0.162 0.1798 -0.0119 0.0325 0.9626 0.9592 0.9415 0.0308 0.0166 0.13 0.0416 0.1628 -0.029 -0.049 0.0094 -0.0446 -0.0405 0.1083 0.097 -0.3638 0.171 0.1344 0.2566 Four -0.0407 -0.0654 -0.0646 -0.0682 -0.074 0.0352 -0.1307 -0.0871 -0.0733 0.1058 -0.1416 -0.0574 -0.1971 -0.0391 -0.0452 0.0666 -0.1197 -0.2006 0.0058 0.0805 0.0767 0.0733 0.8138 0.8084 0.8068 0.7271 0.0442 0.2364 0.1948 -0.0589 -0.0941 0.0241 0.0298 0.5231 0.3071 0.182 0.4082 0.0617 Five -0.0167 -0.0128 0.0069 -0.0778 -0.1036 0.1088 -0.047 -0.1296 0.0181 0.0946 -0.1602 0.0514 -0.121 0.0432 -0.0724 -0.1362 0.2447 -0.1462 -0.1086 0.0792 0.074 0.0717 0.3159 0.2512 0.0963 -0.2703 0.8093 0.7203 0.5932 0.1463 -0.1361 -0.0013 0.009 0.3583 0.277 0.4978 0.3505 0.3173 Six 0.0155 -0.0606 -0.0085 0.0066 -0.0381 0.0991 0.0101 -0.004 0.0058 0.1189 0.0441 -0.0836 -0.0066 0.0218 -0.0002 0.1735 0.0296 0.136 0.5511 -0.0029 0.0154 -0.006 -0.1895 -0.2196 0.1393 0.2358 0.0163 0.0351 -0.0991 0.7999 0.7657 0.2081 0.0531 -0.1013 0.3698 0.4534 -0.1311 0.0115 Seven 0.0282 -0.0615 0.0055 -0.0034 0.035 -0.0176 -0.0421 0.0384 -0.0754 0.0068 -0.0074 0.0004 -0.0742 -0.1468 -0.1644 0.0356 -0.0312 -0.0931 0.1292 0.0124 0.0212 0.0061 0.2624 0.0586 -0.1933 -0.0389 -0.1004 0.2378 -0.2024 0.0061 0.2252 0.8262 0.0189 0.1299 -0.5005 -0.1945 0.5452 0.1648 Eight -0.0275 0.0867 -0.1262 -0.0137 -0.0819 0.0904 -0.0132 -0.0834 0.064 0.0177 -0.2079 0.0355 -0.1419 0.4013 0.4196 0.0114 0.0128 -0.0352 -0.0382 0.0393 0.0374 0.0354 0.0649 0.0351 -0.0995 -0.0596 0.0335 -0.0249 -0.1783 0.0337 0.0728 -0.0949 0.7542 0.258 -0.2068 0.1839 0.1675 0.0205

(Cells show correlation values for r; boxes show correlations where r 2 > .3).

120

represents areas which have high levels of violent crime and which are vulnerable to economic change because the residents are not employed in diverse sectors of the economy. At the tract level, the description for the factor can be a more precise “violent crime and blight” characterization, as nonviolent crimes are somewhat less part of the picture. Factor 5 at the tract level combines some of the characteristics of several of the block group level factors. The tract level factor has favorable associations with Public Assistance (r = .593, r2 = .352) and Vacancies (r = .720, r2 = .518) and an unfavorable association with the Home Affordability indicator (r = .809, r2 = .654). The factor seems to be best interpretable as a “slum” factor. Figure 10 shows the spatial distribution of vacancies in New Orleans. Factors 6 through 8 are largely negligible factors at the tract level, accounting together for a mere 10.4% of the indicator variance. Each of these factors essentially only correlates with one indicator. Factor 6 is an “employment” factor, correlating highly with the Employment and Real Employment indicators. Factor 7 is a “walking” factor, correlating highly with the Walk To Work indicator. Factor 8 is a “doctor’s office” factor, correlating highly with the point-in-polygon Doctors indicator measure. The neighborhood level factor analysis also generated 8 factors, this time accounting for 83.4% of the variance. The first three factors counted for 58.6% of the variance, slightly higher than at the tract level. The correlation matrix for the

121

Figure 10. Percentage of Vacant Units in New Orleans by Census Block Group (quintile distribution)

122

neighborhood level is presented as Table 11. Once again, the interpretations of the first three factors were more or less identical to what they had been previously. The interpretation of Factor 4 was also consistent with its interpretation at the tract level. Factor 5 at the neighborhood level showed a very strong association with Nonviolent Crime (r = .814, r2 = .663). This association had not been present at this strength at either the block group or the tract level, though it will be recalled that the Nonviolent Crime indicator had dropped out of Factor 4 at the tract level. Factor 5 at the neighborhood level appears to be a “slum” factor where petty property crimes play more of a role. Factors 6 through 8 explain only 9.8% of the variance at the neighborhood level and seem even less substantively meaningful than they had at the tract level. Factor 6 is associated favorably with Median Household Income and unfavorably with RTA To Work; it seems to be a “low bus ridership” factor. Factor 7 does not correlate particularly highly with any of the indicators and is more or less a general factor; its most significant characteristic is a moderately favorable association with proximity to fire stations. Factor 8 is a “distance from parks” factor not unlike Factor 6 at the block group level. The “distance to services,” “economic” and “environmental” groups were constant features at all three levels of spatial aggregation. These results verified many of the assumptions of H2, although it was clear there were many important deviations from the assumptions as originally stated.

123

Table 11. Neighborhood Level Correlation Matrix from Factor Analysis
Indicators Schools C lothing LEAP Elementary D octors (C entroid) Libraries Police Stations R ec Facilities G roceries (C entroid) LEAP Junior/Middle C AT Elementary G rad Exam H ospitals Parks (C entroid) Employment R eal Employment Public Assistance H igh School Post H igh School Median Income Violent/N onproperty H ealth Risk T oxic W ater R eleases T oxic R eleases Employment D iversity T raffic Index Violent C rime W alk T o W ork N onviolent C rime Bike T o W ork Vacancies R T A T o W ork Median Value Parks (Perimeter) D octors (Points) G roceries (Points) H ome Affordability C ommute W ave Fire Stations O ne 0.9563 0.9412 0.9334 0.932 0.9128 0.9041 0.8531 0.8312 0.8252 0.8222 0.8048 0.7648 0.7037 0.0533 0.0156 -0.0335 -0.0217 -0.042 0.0679 -0.1579 -0.1906 -0.2171 -0.3412 -0.0249 -0.0807 -0.1678 -0.1049 -0.2054 -0.1127 0.0785 -0.1732 0.0378 -0.0971 -0.047 -0.1795 -0.0234 0.1563 0.3999 T wo 0.0275 -0.0321 -0.0116 -0.0317 -0.0451 0.124 0.0476 0.1391 0.061 -0.0871 -0.0223 0.0645 -0.0626 0.9472 0.9452 -0.9395 0.8808 0.7494 0.6062 -0.5815 0.0371 0.0489 0.0351 0.2245 0.208 -0.4218 0.1378 -0.0673 0.2542 -0.2988 -0.13 0.5016 0.06 0.1402 -0.1474 -0.3042 -0.2598 0.0686 T hree -0.0791 -0.096 -0.0928 -0.0497 -0.2218 0.0456 -0.0877 -0.0677 -0.2211 -0.0456 -0.4132 -0.0752 -0.1202 0.0482 -0.045 0.0046 0.0329 0.2187 0.0852 0.0034 0.9543 0.9519 0.9195 0.0038 0.0942 0.0173 -0.0018 0.1163 0.2733 -0.0374 0.2673 0.3412 -0.007 0.1792 0.1572 0.3923 -0.4912 0.0131 Four -0.0404 -0.0751 -0.0044 -0.0727 -0.0186 -0.1188 -0.0448 -0.0827 -0.0759 0.1203 -0.0861 -0.0567 0.0226 -0.1162 0.0228 -0.0609 0.1564 0.2396 -0.2326 0.7212 0.0788 0.0732 0.0662 0.9186 0.8985 0.738 0.0661 0.1817 -0.0223 -0.211 -0.0685 -0.2257 -0.0186 0.0195 0.4708 -0.2711 0.3492 -0.293 Five -0.0453 -0.0949 -0.0123 -0.0289 -0.124 -0.0149 -0.1559 -0.1094 0.072 0.0272 -0.0859 -0.2174 -0.1324 -0.0161 0.0886 0.071 0.057 0.1259 -0.1009 0.2262 0.0942 0.0881 0.0873 -0.0079 -0.1309 0.4495 0.8621 0.8137 0.6971 0.6354 0.2161 0.1908 -0.1324 -0.0029 0.5309 0.4308 -0.3526 -0.1987 Six 0.0621 0.0528 -0.0626 -0.012 0.0333 0.0174 0.1792 0.1529 -0.0106 -0.2289 0.0247 0.2164 -0.0236 -0.0801 -0.0835 -0.1443 0.3663 0.4768 0.6402 0.0286 -0.0443 -0.0323 -0.0375 0.0242 -0.0391 -0.0207 0.1374 -0.1099 -0.0478 -0.3828 -0.8018 0.5492 -0.0866 -0.1452 -0.398 -0.0283 -0.1207 0.0973 Seven -0.0312 0.0098 0.1222 -0.1994 -0.0448 -0.0388 -0.1022 -0.2958 0.0371 0.0341 0.1729 -0.1484 0.4853 0.0548 -0.0957 0.0307 -0.0096 0.0583 -0.0408 0.0718 0.0105 0.0072 -0.0145 -0.0176 0.0706 0.0629 -0.0536 0.0728 -0.0548 0.19 0.1608 0.176 0.065 0.4865 0.0932 0.3563 -0.2098 -0.5403 Eight -0.0079 -0.0352 -0.0231 0.0472 -0.1026 -0.0269 -0.0522 -0.0382 -0.0873 0.0328 -0.0023 0.1038 0.16 -0.0561 -0.0429 0.0153 0.0015 -0.0888 -0.0024 -0.0257 -0.0432 -0.0211 -0.042 0.0318 -0.0708 -0.027 -0.087 -0.0541 -0.0561 -0.0551 0.0234 0.0147 0.858 -0.5435 0.1779 0.083 0.0404 0.0217

(Cells show correlation values for r; boxes show correlations where r 2 > .3).

124

Regressions on the Indicator Data In order to gain insight into the benefit home ownership has in terms of quality of life in the most general sense, an index of the 38 indicators was constructed. This index was a sum of standardized scores (or “z-scores”) for each of the indicators. The z-scores were added into the index in the direction of favorability for quality of life. In other words, if a positive value for an indicator meant higher quality of life, z-scores for that indicator were added into the index, while if a negative value for the indicator meant a higher quality of life, z-scores for that indicator were subtracted from the index. This 38indicator index was given the name “Index 1.” A regression of percentage home ownership on Index 1 was attempted at all three levels of spatial aggregation. Analysis of variance diagnostics on the regressions revealed that the results of all three regressions were insignificant. Collinearity with the Median Household Income indicator seems to have been responsible for this outcome. When a new 36-indicator index (named “Index 2”) was constructed eliminating the Median Household Income and Home Affordability indicators (the latter indicator was eliminated because Median Household Income is used to calculate it), significant relationships were discovered at the first two levels of aggregation. In Table 12 the standardized regression coefficients or “beta weights” (b) for home ownership and median household income with Index 2 are presented for all three levels of aggregation. At the block group and tract levels, median household income is shown to have a mild but positive association with the index, while home ownership has

125

a mild but importantly negative association with the index. At the neighborhood level, the beta weight for home ownership is not statistically significant. This means that the regression model cannot establish a linear relationship between home ownership and Index 2 at the neighborhood level.
Table 12. Values for b of Percentage Home Ownership and Median Household Income with the 36-Indicator Quality of Life Index Block Group Tract Neighborhood Hom e Ownership -.310** -.453** -.255 Median Household Incom e .356** .440** .412* * Significant at .05 level. ** Significant at .01 level.

For this reason, to gain further insight into the relationship between home ownership and the indicators at the neighborhood level, a curvilinear regression model was employed. The specific model, using a cubic polynomial function to fit a curve to the data, has been described at the end of the statistical methodology section of this thesis. Using the cubic polynomial model, it was possible to identify a rather interesting relation home ownership bears to the 36-indicator index only at the neighborhood level. As Figure 11 on the following page shows, a curve fit to the home ownership data at the neighborhood level reveals that home ownership causes the index to rise up to a point around a home ownership level of 30%. At that point, home ownership actually begins to cause a decline in the index. This decline continues until home ownership rates

126

Figure 11. Cubic Polynomial Regression of Home Ownership on the 36-Indicator Quality of Life Index at the City Neighborhood Level

127

reach around 70%, at which point home ownership once again begins to positively affect the index. This is not a minor decline. The decline represents a change in the z-score index from a positive ranking of .2 at the first “turning point” to a ranking of just under zero at the second “turning point.” This means that at around 30% home ownership, the quality of life indicators improve slightly, whereas around 70% home ownership they do not do anything at all. This is not a “leveling off” of the relationship; it constitutes a significant drop. Since this is a graph of a third-order polynomial function, there can be no more than two “turning points” on the graph, because there can never be more “turning points” than the order of the polynomial minus one. It is possible that a higher order polynomial function could have produced more “turning points” on the graph of this relationship than the two which it does in fact show. These two are sufficient, however, to point out an important difference between the regression of home ownership on the index and the regression of median household income on the index. A curve for a simple cubic function, like y = x3, will be an inverse curve with no “turning points.” What is important about the above curve is that it is not an inverse curve, but a curve that does have turning points. It is by no means a necessary feature of a cubic polynomial function that it appear like the regression shown in Figure 11. Indeed, the graph for the regression of median household income on Index 2 appears to be much closer to an inverse curve of the kind provided by a simple cubic function, rather

128

than a curve with strongly identifiable “turning points.” Figure 12 is the graph of this regression. There is clearly much less of an oscillation in the median household income data, which is why a linear relationship could be approximated within significance requirements in linear regression for median household income even at the neighborhood level. The strong oscillation within the home ownership data, by contrast, necessarily yielded linear approximations of the relationship insignificant. Importantly, once median household income rises above roughly $15,000, the index always remains above zero, which means that after that point income always affects the index and does so positively. This is a strong contrast with home ownership, which returns to zero at around 70% home ownership. The reason for the great oscillation in the neighborhood level relationship between home ownership and the index may have something to do with homogeneity of tenure. Home ownership seems to have a negative relation to the index in areas where home owners and renters live together in roughly equal numbers. A possible interpretation is that conflict between the two tenures is responsible for a downturn in quality of life. This would explain why this relationship became apparent only at the most expansive spatial level of aggregation; neighborhoods are more likely to group together diverse populations and to be fractured sociologically along any number of lines.

129

Figure 12. Cubic Polynomial Regression of Median Household Income on the 36-Indicator Quality of Life Index at the City Neigborhood Level

130

Comparisons of the impact of home ownership and median household income on quality of life can also be carried over to specific factors from the factor analysis. A similar method of analysis to the one used with the quality of life indices was employed to get at the relative importance of home ownership and median household income with regard to certain groups of indicators. Factor analysis was run on 36 indicators (once again eliminating the Median Household Income and Home Affordability indicators) yielding factors very similar to the original factors but without any contribution from median household income. There were fewer factors in these new analyses (9 at the block group level, 8 at the tract level and 7 at the neighborhood level) but the first three factors were clearly the same “distance to services,” “economic” and “environmental” factors which appeared in the earlier analyses. A comparison of home ownership’s impact on these three factors relative to the impact of median household income yields some interesting results. Table 13 shows the correlation matrix for the factors against these two variables.
Table 13. Correlations of Home Ownership and Median Household Income with Major Factors Block Group Tract Neighborhood Home Ow nership "Distance To Services" .196** .235** .084 "Econom ic" .492** .512** .562** "Environm ental" -.255** -.127* -.137 Median Household Income "Distance To Services" .112** .077 .070 "Econom ic" .717** .610** .648** "Environm ental" .035 .036 .103

131

Two things can be gleaned from the preceding table. First, both home ownership and median household income affect the “economic” category of indicators far more than any other of the categories. Second, home ownership affects the factors more negatively than median household income. Home ownership is associated with greater distance from services and with less environmental protection than median household income, and is not related nearly as strongly to the “economic” factor as is income. The moderately high correlation with a factor representing lack of services is fairly disturbing. Individuals are, by current city policies, being encouraged to purchase homes, but areas with a high percentage of home ownership are not proximate to services needed for the operation of those homes. This is contradictory in the extreme, and the results of such contradictions can be seen in every day life. “Lack of services for home operation,” suggest Matthew Edel, Elliott Sclar and Daniel Luria, has in the past “led ‘stable citizens’ to hurl snowballs at mayors and deposit garbage on city hall steps.” (Edel, Sclar and Luria, 1984: 6). This is no small matter in people’s lives, and should be treated as the important issue it is. This suggests that, if home ownership is being touted as a means of neighborhood revitalization, there are more useful means available. There is little here to strongly recommend home ownership as a means of increasing the quality of life in lower-income communities. Indeed, it would seem from the above that a more-to-the-point strategy would be replacing finding a way to get lower-income communities some income.

132

CHAPTER VIII CONCLUSIONS

This chapter is divided into two sections. The first section outlines the theoretical implications of the findings of this thesis for geography and housing and community development. The second section exposits the policy ramifications of the principal findings of the thesis.

Theory Implications The theoretical findings of this thesis can be summarized as follows: · Home ownership’s main contribution to quality of life is economic. The quality of life indicators most strongly benefited by home ownership are indicators which are expressly economic, and non-economic indicators are either not affected or are actually negatively affected by increased levels of home ownership. · Home ownership does not affect quality of life more strongly than income level. Even the economic indicators are not affected by home ownership to the extent that they are affected by income levels. · The “three E’s” model of indicator domains advanced by the urban sustainability paradigm, though imperfect, seem to approach the domains one

133

actually finds empirically. Though more specific interpretations of the indicator domains which resulted from factor analysis are possible, these interpretations do not vary tremendously from what was expected by the theory. · An increase in geographic scale intensifies relationships rather than varying them. Contrary to expectations, different domains of indicators do not come into play at different levels of spatial aggregation. Relationships between home ownership and indicator domains tend to increase in clarity, with positive relationships becoming more positive and negative relationships becoming more negative.

Policy Implications The implications for future policy at which this thesis arrives are as follows: · Rather than assume that home ownership is a scheme for neighborhood revitalization superior to all others, evaluation of alternative tenure policies should be considered. The findings of this thesis suggest that home ownership is one of many things which may play a role in increasing the quality of life, but does not present striking evidence by any means of the superiority of home ownership strategies of revitalization. The option of home ownership will likely always be a valued option in the United States,

134

and certainly government should do nothing to enjoin individuals from owning their own homes. But if assistance is offered, the rationale for this should probably be to expand consumer choice rather than to create a kind of picket fence Utopia. · An effort should be marshaled to introduce needed services to areas which are remote from them. In New Orleans, the predominant indicator factor represents what is perhaps the city’s most obvious design flaw – people do not live close to services they wish to utilize on a regular basis, some of which are even necessary services. Home ownership areas are not immune from this problem, and indeed, increased levels of home ownership seem to mildly aggravate the problem. Perhaps most serious is the distance separating people from hospitals to which they would turn in a pressing medical emergency situation. The data indicates that, paradoxically, the most affluent area in the city suffers the most greatly from being cut off from conveniences and necessities. Furthermore, failure to provide for the delivery of needed services to home owners is inconsistent with an overall policy which encourages home ownership. · The encouragement of lower-income individuals to own homes via mortgage rate discounting programs will not help raise the quality of life in lowerincome areas unless rising property values allow these individuals to raise

135

their incomes by saving a great deal of equity. The numbers indicate that income is more likely to have an effect on quality of life than ownership tenure per se. Even income is not a sole determinant of quality of life by any means, but there is little sense in encouraging low-income individuals to own their own homes without a prospect for retaining equity in the property and using it to bolster their economic resources. Mortgage assistance programs should consider themselves as making an implicit promise of equity to clients, as without it, the claim that an individual is bettering his life by purchasing a home is problematic. There is little connection between tenure choice and quality of life, and should the bottom fall out of property values in an area, the low-income home owner would be in debt, essentially because government experts and non-profit organizations determined that he should purchase rather than rent. Perhaps a monetarist-type control over mortgage assistance might be in order. When property values are likely to be on the rise, more assistance might be offered; when they are likely to decline, the policy might be to pull back on the reins and offer less assistance. · Prejudice expressed in policy against renters is unjustified. The findings of this thesis utterly fail to confirm the suspicion that declining quality of life is the fault of the person who rents rather than owns. The time and effort expended trying to convert renters, as if they were representatives of a heathen

136

tribe of savages, into civilized owners, seems sadly misplaced given the weak support home ownership obtains from this thesis.

137

Bibliography Appleby, Joyce. 1984. Capitalism and a new social order: the republican vision of the 1790s. New York, NY, USA: New York University Press. Babbie, Earl. 1995. The practice of social research. Belmont, CA, USA: Wadsworth. Bauer, Raymond. 1966. Social indicators. Cambridge, MA, USA: MIT Press. Becker, Gary. 1995. The essence of Becker. Edited by Ramón Febrero and Pedro S. Schwartz. Stanford, CA, USA: Hoover Institution Press. Bell, Daniel. 1969. “Towards a social report.” Public Interest 15:72-84. Berry, Brian J. L. 1973. The human consequences of urbanization. New York, NY, USA: St. Martin’s Press. Berry, Brian J. L. and Frank Horton. 1970. Geographic perspectives on urban systems. Englewood Cliffs, NJ, USA: Prentice-Hall. Berry, Brian J. L. and John D. Kasarda. 1977. Contemporary urban ecology. New York, NY, USA: Macmillan. Bolton, Roger. 1992. “Place prosperity versus people prosperity revisited: an old issue with a new angle.” Urban Studies 29(2):185-203. Brown, Lester R. 1981. Building a sustainable society. New York, NY, USA: W. W. Norton and Company. Campbell, Angus. 1981. The sense of well-being in America. New York, NY, USA: McGraw-Hill.
„ apek, Stella and John I. Gilderbloom. 1992. Community versus commodity: tenants

and the American city. Albany, NY, USA: SUNY Press. Center for Sustainable Communities. 1997. Center for Sustainable Communities home page. At http://weber.u.washington.edu/~common/ on the World Wide Web. Seattle, WA, USA: University of Washington.

138

Chatterjee, Samprit and Bertram Price. 1991. Regression analysis by example. New York, NY, USA: John Wiley and Sons. Clark, William A. V. and Eric Moore, eds. Residential mobility and public policy. 1980. Beverly Hills, CA, USA: Sage Publications. Cole H. S., D. Christopher Freeman, Marie Jahoda and K. L. R. Pavitt, eds. 1973. Models of doom: a critique of the limits to growth. New York, NY, USA: Universe Books. Comrey, Andrew L. 1973. A first course in factor analysis. New York, NY, USA: Academic Press. Data Analysis Unit, City of New Orleans. 1983. Neighborhood characteristics summary 1983. New Orleans, LA, USA: City of New Orleans. Donaldson, Scott. 1969. The suburban myth. New York, NY, USA: Columbia University Press. DeLeeuw, Frank and Raymond Struyk. 1975. The web of urban housing. Washington, DC, USA: The Urban Institute. Doucet, Michael and John Weaver. 1991. Housing the North American city. Kingston, ON, Canada: Queen’s University Press. Dreier, Peter. 1982. “The status of tenants in the United States.” Social Problems 30(2, December):179-98. Drummond, William. “The development of GIS-based small area social indicators.” Proceedings of the Third International Conference on Computers in Urban Planning and Urban Management. Edel, Matthew, Elliott Sclar and Daniel Luria. 1984. Shaky palaces: homeownership and social mobility in Boston’s suburbanization. New York, NY, USA: Columbia University Press. Engels, Friedrich. 1975 (1872). “On the housing question.” In Marx and Engels: Collected Works. New York, NY, USA: International Publishers.

139

Entwhistle, Barbara. 1991. “Micro-macro theoretical linkages in social demography: a commentary.” In Micro-macro linkages in sociology, edited by Joan Huber. Newbury Park, CA, USA: Sage Publications. Frank, Joseph. 1955. The Levellers: a history of the writings of three seventeenthcentury social democrats. New York, NY, USA: Russell and Russell. Galster, George C. 1986. “What is neighborhood: an externality-space approach.” International Journal of Urban and Regional Research 10:243-61. Galster, George C. 1987. Homeowners and neighborhood reinvestment. Durham, NC, USA: Duke University Press. Garn, Harvey A., Michael J. Flax, Michael Springer and Jeremy B. Taylor. 1976. Models for indicator development: a framework for policy analysis. Washington, DC, USA: The Urban Institute. George, Henry. 1966 (1879). Progress and poverty. Preface by A. W. Madsen. London, UK: Hogarth Press. Gilderbloom, John I. 1981. “Moderate rent control: its impact on the quality and the quantity of the housing stock.” Urban Affairs Quarterly 17(2):123-42. Goetz, Edward. 1995. “A little pregnant: the impact of rent control on San Francisco.” Urban Affairs 30(4):604-12. Goetze, Rolf. 1976. Building neighborhood confidence. Cambridge, MA, USA: Ballinger. Goetze, Rolf. 1979. Understanding neighborhood change. Cambridge, MA, USA: Ballinger. Goetze, Rolf and Kent Colton. 1980. “The dynamics of neighborhood.” Journal of the American Planning Association 46:184-94. Granovetter, Mark. 1978. “Threshold models of collective behavior.” American Journal of Sociology 83:1420-33.

140

Harrington, James. 1992 (1656). The commonwealth of Oceana. In The commonwealth of Oceana, and a system of politics, edited by James G. A. Pocock. Cambridge, UK: Cambridge University Press. Harrington, James. 1992 (1661). A system of politics: delineated in short and easy aphorisms. In The commonwealth of Oceana, and a system of politics, edited by James G. A. Pocock. Cambridge, UK: Cambridge University Press. Hart, Maureen. 1995. Guide to sustainable community indicators. Ipswich, MA, USA: QLF/Atlantic Center for the Environment. Hartman, Chester and Michael Stone. 1986. A socialist housing alternative. In Critical perspectives in housing, edited by Rachel Bratt, Chester Hartman and Ann Meyerson. Philadelphia, PA, USA: Temple University Press. Harvey, David. 1985. The urbanization of capital: studies in the history and theory of capitalist urbanization. Baltimore, MD, USA: Johns Hopkins University. Hays, R. Allen. 1985. The federal government and urban housing: ideology and change in public policy. Albany, NY, USA: SUNY Press. Heskin, Allan David. 1984. Tenants and the American dream. New York, NY, USA: Praeger. Heskin, Allan David. 1995. The struggle for community. Boulder, CO, USA: Westview Press. Higgins, Benjamin and Jean Downing Higgins. 1979. Economic development of a small planet. New York, NY, USA: W. W. Norton and Company. Hopkins, Terence. 1982. World-systems analysis: theory and methodology. Beverly Hills, CA, USA: Sage Publications. Hotelling, H. 1933. “Analysis of a complex of statistical variables into principal components.” Journal of Educational Psychology 24:417-41, 498-520. Jacksonville Community Council, Incorporated. 1997 [1996.] Building community in the American tradition. At http://libertynet.org/~edcivic/jacksonv.html on the World Wide Web. Jacksonville, FL, USA: Jacksonville Community Council, Incorporated.

141

Jacobs, Jane Margaret. 1984. Cities and the wealth of nations: principles of economic life. New York, NY, USA: Random House. Jefferson, Thomas. 1984 [1785.] Letter to John Adams. In Jefferson: Writings. Washington, DC, USA: Library of America. Jones, Stephen R. G. 1984. The economics of conformism. Cambridge, MA, USA: Blackwell. Kahn, Herman, William Brown and Leon Martel. 1976. The next 200 years: a scenario for America and the world. New York, NY, USA: William Morrow and Company. Kain, John F. and William Apgar. 1985. Housing and neighborhood dynamics: a simulation study. Cambridge, MA, USA: Harvard University Press. Kemeny, Jim. 1981. The myth of home ownership. London, UK: Routledge and Kegan. Kemp, Jack. 1990. Secretary Jack Kemp offers a progressive-conservative prescription for a new war on poverty. Washington, DC, USA: Department of Housing and Urban Development. Kim, S. B. 1978. Landlord and tenant in colonial New York. Chapel Hill, NC, USA: University of North Carolina Press. King, Gary. 1986. “How not to lie with statistics: avoiding common mistakes in quantitative political science.” American Journal of Political Science 30(3):66687. Leccese, Michael. 1997. “Ship shape.” Landscape Architecture. 87(9):34-39. Lial, Margaret L., E. John Hornsby and David I. Schneider. 1997. College algebra and trigonometry. Reading, MA, USA: Addison-Wesley. Locke, John. 1967 (1699). Two treatises on government. Edited by Peter Laslett. London, UK: Cambridge University Press.

142

Lorr, Maurice. 1983. Cluster analysis for social scientists: techniques for analyzing and simplifying complex blocks of data. San Francisco, CA, USA: Jossey-Bass Publishers. Luger, Michael I. 1996. “Quality of life differences and urban and regional outcomes.” Housing Policy Debate 7(4):749-71. Lynch, Kevin. 1960. The image of the city. Cambridge, MA, USA: MIT Press. Macpherson, Crawford Brough. 1978. Property: mainstream and critical positions. Toronto, ON, Canada: University of Toronto Press. Marcuse, Peter. 1971. “Social indicators and housing policy.” Urban Affairs Quarterly 7:193-217. Marcuse, Peter. 1986. “The myth of the benevolent state.” In Critical perspectives in housing, edited by Rachel Bratt, Chester Hartman and Ann Meyerson. Philadelphia, PA, USA: Temple University Press. Marx, Karl. 1957 (1867). Capital. New York, NY, USA: Dutton. Marx, Karl. 1976 (1845). “On the German ideology.” In Marx and Engels: collected works. Moscow, Russia: International Publishers. Massey, Douglas S. and Nancy A. Denton. 1993. American apartheid. Cambridge, MA, USA: Harvard University Press. McHarg, Ian. 1969. Design with nature. Garden City, NY, USA: The Natural History Press. Meadows, Donella H., Dennis L. Meadows, Jørgen Randers and William W. Behrens III. 1972. The limits to growth. New York, NY, USA: Universe Books. Michaelson, William. 1977. Environmental choice, human behavior and residential satisfaction. New York, NY, USA: Oxford University Press. Milgram, S., J. Greenwald, S. Kessler, W. McKenna and J. Waters. 1972. “A psychological map of New York City.” American Scientist 60: 194-200. Mishan, E. J. 1967. The costs of economic growth. London, UK: Staples Press.

143

Morris, Earl, Sue Crull and Mary Winter. 1976. “Housing norms, housing satisfaction and the propensity to move.” Journal of Marriage and the Family 38:309-20. Myers, Dowell. 1987. “Community-relevant measurement of quality of life.” Urban Affairs Quarterly 23:108-25. Myers, Dowell. 1988. “Building knowledge about quality of life for urban planning.” Journal of the American Planning Association 54:347-58. National Association of Home Builders. 1997. The HOME page. At http://www.nahb.com on the World Wide Web. Washington, DC, USA: National Association of Home Builders. Norušis, Marija. 1990. SPSS advanced statistics student guide. Chicago, IL, USA: SPSS. Office of Policy Planning and Analysis, City of New Orleans. 1978. An analysis of blight in the city of New Orleans: blight index, 1976. New Orleans, LA, USA: City of New Orleans. Ozanne, Larry and Raymond Struyk. 1976. Housing from the existing stock. Washington, DC, USA: The Urban Institute. Parsons, Robert. 1974. Statistical analysis: a decision-making approach. New York, NY, USA: Harper and Row. Perin, Constance. 1978. Everything in its place. Princeton, NJ, USA: Princeton University Press. Piven, Frances Fox and Richard Cloward. 1974 (1967). “Rent strike: disrupting the slum system.” In The politics of turmoil. New York, NY, USA: Pantheon. Raspberry, William. 1990. “Kemp at HUD.” Raleigh News and Observer, 15 May. Ricardo, David. 1970 (1817). Works of David Ricardo, volume one: principles of political economy. Edited by Piero Sraffa. Cambridge, UK: Cambridge University Press. Rice, Stuart Arthur. 1969. Farmers and workers in American politics. New York, NY, USA: AMS Press.

144

Rohe, William A. and Michael A. Stegman. 1992. “Public housing homeownership: will it work and for whom?” Journal of the American Planning Association 58(2, Spring):144-57. Rosenthal, Donald B. 1988. Urban housing and neighborhood revitalization. New York, NY, USA: Greenwood Press. Rummel, R. J. 1970. Applied factor analysis. Evanston, IL, USA: Northwestern University Press. Sawicki, David and Patrice Flynn. 1996. “Neighborhood indicators: a review of the literature and an assessment of conceptual and methodological issues.” Journal of the American Planning Association 62(2, Spring): 165-83. Sayer, Andrew J. 1984. Method in social science: a realist approach. London, UK: Hutchinson. Schlesinger, Arthur. 1945. The age of Jackson. Boston, MA, USA: Little, Brown and Company. Schoenberg, Sandra and Patricia Rosenbaum. 1980. Neighborhoods that work. New Brunswick, NJ, USA: Rutgers University Press. Sellers, Charles. 1991. The market revolution. New York, NY, USA: Oxford University Press. Shipley, Sara. 1997. “The best schools.” New Orleans Times-Picayune. 6 July. Page A1. Smith, Wallace. 1970. Housing: the social and economic elements. Berkeley, CA, USA: University of California Press. Spirn, Anne Whiston. 1984. The granite garden: urban nature and human design. New York, NY, USA: Basic Books. Stone, Michael. 1971. “Housing, mortgages and the state.” Upstart (3, December):2334.

145

Straussfogel, Debra. 1997. “Redefining development as humane and sustainable.” Annals of the Association of American Geographers 87 (2, June):280-305. Sustainable Seattle. 1997 (1994). Sustainable Seattle. At http://www.scn.org/scripts/menus/envtech/sustainable/Sustainable.html on the World Wide Web. Seattle, WA, USA: Sustainable Seattle. Sweeney, James. 1974. “Housing unit maintenance and the mode of tenure.” Journal of Economic Theory 8:111-38. Taggart, Robert. 1970. Low income housing: a critique of federal aid. Baltimore, MD, USA: Johns Hopkins University Press. Talen, Emily. 1997. “After the plans: methods to evaluate the implementation success of plans.” Journal of Planning Education and Research 16(2, Winter):79-91. Taylor, John. 1818. Arator, being a series of agricultural essays, practical and political. Petersburg, VA, USA: Whitworth and Yancey. Thurstone, L. L. 1947. Multiple-factor analysis. Chicago, IL, USA: Chicago University Press. Tremblay, Kenneth and Don Dillman. 1983. Beyond the American housing dream: accomodation to the 1980s. Lanham, MD, USA: University Press of America. Turner, Frederick Jackson. 1920. The frontier in American history. New York, NY, USA: H. Holt and Company. United States Department of the Census. 1997 (1990). 1990 Census Lookup. At http://venus.census.gov/cdrom/lookup/ on the World Wide Web. Washington, DC, USA: Department of the Census. United States Department of the Census. 1997 (1994). Tiger Map Server. At http://tiger.census.gov/ on the World Wide Web. Washington, DC, USA: Department of the Census. United States Environmental Protection Agency, 1995 Toxic Release Inventory. At http://www.rtk.net on the World Wide Web. Washington, DC, USA: Environmental Protection Agency.

146

Urban Quality Indicators. 1997. Edited by Cy Yoakam. Volume 1, Issues 1-4. Ann Arbor, MI, USA: Urban Quality Communications. Varady, David. 1986. Neighborhood upgrading: a realistic assessment. Albany, NY, USA: SUNY Press. Wallerstein, Immanuel Maurice. 1991. Geopolitics and geoculture: essays on the changing world-system. Cambridge, UK: Cambridge University Press. Weber, Max. 1985 (1905). The protestant ethic and the spirit of capitalism. New York, NY, USA: Charles Scribner’s Sons. Wilson, Alan Geoffrey. 1974. Urban and regional models in geography and planning. New York, NY, USA: John Wiley and Sons. Wilson, William Julius. 1987. The truly disadvantaged. Chicago, IL, USA: University of Chicago Press. Wish, Naomi Bailin. 1986. “Are we really measuring quality of life?” American Journal of Sociology 45:93-99. Wood, Robert C. 1958. Suburbia: its people and its politics. Boston, MA, USA: Houghton Mifflin and Company. Wright, Gwendolyn. 1981. Building the dream. New York, NY, USA: Pantheon. Younger, Mary Sue. 1979. Handbook for linear regression. North Scituate, TN, USA: Duxbury Press. Zundel, Alan. 1995. “Policy frames and ethical traditions: the case for home ownership for the poor.” Policy Studies Journal 23(3):423-34.

147

Vita Zachary Klaas was born in Bellflower, California, on 24 December 1965. He was raised in Ames, Iowa, and graduated from Iowa State University in Ames with a Bachelor of Arts in History and Political Science in May, 1990. He moved to New Orleans, Louisiana, in June, 1990, and enrolled at the University of New Orleans, from which he graduated with a Bachelor of Arts degree in Philosophy in May of 1992. During his time in New Orleans he worked at a pair of art house movie theatres, the Canal Place and Prytania theatres, and at a Lakefront store in the venerable Schwegmann’s chain of grocery stores. He entered the Master of Urban and Regional Planning program at the University of New Orleans in August of 1994 and with this master’s thesis completes that program. During his time in the planning program, he concentrated in geographic information systems technology, and hopes to find employment in a geography-related field. In October, 1997 he moved to Ottawa, Ontario, Canada, where he is currently residing and actively seeking employment in planning and geographic information systems fields.