Research Methodology

Published on July 2016 | Categories: Types, Presentations | Downloads: 34 | Comments: 0 | Views: 193
of 25
Download PDF   Embed   Report

Research Methodology notes

Comments

Content


1
NOTES ON RESEARCH METHODS
Michael Wood (email: [email protected] )
Portsmouth University Business School
Contents
Introduction to research methodology …
Strategies or research pro!ects
Research aims or !uestions … "
#eneral issues concerning research$ %hiloso%hy& etc … '
Understanding the present" predicting the uture" and improving the uture
Positivism and phenomenology" and similar distinctions
#he degree o generality
#heories: $uilding" testing" amending" using
(olitics and ethics … )
Research design … )
%mpirical methods
Surveys
%&periments and 'uasi(e&periments
)ase studies and small sample research
*ction research
+odelling
* general design or a typical +asters degree pro!ect
,inking methods to research aims or 'uestions
Data collection methods … "
-nterviews
.uestionnaires
Sampling
Trust*orthiness … +
/alidity
0elia$ility
1$!ectivity
#riangulation
Statistical hypothesis tests
Data analysis … )
#ypes o measurement
)omputer sotware
Writing the re%ort … ,
#he critical attitude
Pu$lishing your research
Chec-list *hen starting a %ro.ect /// and 0inishing it … "1
Re0erences … "1
A%%endices … ""
* note on 2theory2
%&ample to show analysis o 'uestionnaire data
Introduction to research methodology
#his is an area where there is considera$le disagreement on the deinition o concepts" and what is
right and wrong. *ccordingly you should read widely and critically3 never assume that you need
to accept every concept and every assertion. 4ou will pro$a$ly $e a$le to ind an e&ception to
5
every rule (see 6eyera$end" 1789" or an e&treme version o this principle).
#hese notes are intended as a $rie overview o the main issues. -t is important that you
read in more depth on the speciic issues o particular concern to you. 6or e&ample" i you intend
to conduct some interviews or a 'uestionnaire survey" it is important that you consult a suita$le
source o guidance on surveys" interviews and 'uestionnaires : eg Saunders et al (5;;<)" 0o$son
(5;;5)" %aster$y(Smith et al (5;;5).
- will use the word 2method2 or a speciic research method such as a 'uestionnaire
survey. #he word 2methodology2 reers to the study o methods in the same way as 2psychology2
is the study o the psyche. * research 2strategy2 is the overall approach to the pro!ect ( which may
include the use o several methods.
#he word 2research2 in this conte&t covers everything that academic researchers do: the
gathering o inormation a$out the world" the discovery and creation o theories and models to
make sense o this inormation" reviewing and collating research done $y others" as well as
conceptual" mathematical and computational analysis.
Strategies 0or research %ro.ects
#he strategy or carrying out a research pro!ect is largely a matter o common sense. -t is
important not to let !argon and technicalities o$scure this. (- am using the term strategy here in
the sense o a general answer to the 'uestion 2=ow do - go a$out research>2 ( taking all aspects
into account. 4ou will ind other authors may use the term in a slightly dierent sense.)
* simple $asic strategy or any research pro!ect is:
1 ?ecide what you want to achieve ( the aims o the pro!ect" or the 'uestions it will answer.
5 ?ecide how you are going to achieve these aims or answer these 'uestions ( the design o
your research pro!ect. (+ost aspects o research tend to take longer than anticipated" so it
is important to plan the timescale careully to take this into account.)
< )arry out the research" analyse the results and state the conclusions and (i appropriate)
recommendations.
@ )heck that you have in act achieved the aims o the pro!ect. - you have not" work out
your e&cuses" try again" or pretend that you were really trying to do something else ( ie
change your aims to it what you actually did.
1ne diiculty with this is that you may not know e&actly what you want to achieve at the outset.
#his may only $ecome clear as the research progresses. Similarly the appropriate methods (step
5) may only $ecome clear as the research evolves. -n general" it is $est to plan your research in
advance as ar as possi$le" $ut it is clearly important to $e le&i$le.
Research aims or !uestions
Sometimes the research aims or 'uestions are 'uite clear. +ore typically" a research pro!ect may
start rom a airly uAAy pro$lem or area o concern3 it is then necessary to decide on a clear ocus
$y ormulating some more deinite aims or 'uestions ( although you may change your mind a$out
these as discussed a$ove. #his process o achieving a ocus is oten not easy and deserves care
(see Saunders et al" 5;;<" )hapter 5). It is almost always better to focus on a limited area so
that you can do a thorough job, rather than having a broad focus with inevitably superficial
results.
-t is normal to include a section on the $ackground conte&t o the research pro!ect. *s
well as details o the real world issues the pro!ect tackles" you may also wish to discuss the
academic $ackground and your personal motivation. (4our personal aims or doing the pro!ect (
perhaps to pass the course and ac'uire a marketa$le skill ( are" o course" distinct rom the
research aims o the pro!ect.)
#he ocus or your research pro!ect" its goals" can then $e ormulated in any o the
<
ollowing ways:
B .uestion(s) to $e answered: eg What is the best quality strategy for ABC Company
B *im(s) (or o$!ectives) to the achieved: eg !o devise the best quality strategy for ABC
Company.
B * hypothesis or hypotheses to $e tested: eg "trategy # is the best strategy for ABC
Company.
*ims and 'uestions are more or less e'uivalent. Chether you e&press your goals as a list o aims
or as a series o 'uestions does not matter much.
+y preerence would $e or 'uestions $ecause 'uestions lead to answers which can $e
written down in a research report" whereas aims may $e wider than this. 6or e&ample" the aim 2to
increase proits2 is not an appropriate aim or a research pro!ect $ecause the output is not
research. #his is a $usiness aim not a research aim. #he corresponding research aim would $e to
find out how $est to increase proits. 1n the other hand" Saunders et al (5;;<) recommend
o$!ectives $ecause they 2lead to greater speciicity2 (p. 59).
$owever, in general, I would advise you against formulating the aims of your project as
a series of hypotheses to be tested. #esting hypotheses in management is more diicult than it
may appear" and the results o the research $ecome a simple list o #rueD6alse statements ( which
may $e $oring or readersE
?espite this" it may $e useul to have an informal hypothesis ( eg #.+ is helpul ( to
guide your research. #hen you can ormulate some more detailed aims spelling out which aspects
o the helpulness o #.+ that you wish to investigate.
4ou may also have hypotheses you wish to test as a part o addressing your research
aims. 6or e&ample" you may wish to test the hypothesis that there is no dierence in eectiveness
$etween two procedures.
-t is oten helpul to have a series o 'uestions (or aims)" which may $e $roken down into
a hierarchy ( or e&ample:
1
#his diagram shows the airly vague topic 2Strategy to improve F in organisation 42 $roken
@
down into three more speciic o$!ectives. #his is a typical general aim or a +asters degree
pro!ect: F might stand or 'uality" proita$ility" marketing or employee !o$ satisaction" or
e&ample. %ach o these o$!ectives is then applied to two areas o the organisation. #here may $e
more areas to consider" $ut the diagram indicates that this pro!ect is only concerned with two o
them.
* diagram such as this ($ased on Geeney" 1775) should $e helpul or clariying and
structuring your aims (or o$!ectives" or 'uestions). -t is also helpul or checking that your
proposed research methods are likely to $e ade'uate or meeting your aims (or answering your
research 'uestions). CeHll return to this $elow.
#he research aims or 'uestions should
% be unambiguous and clear&
% be coherent, and reasonably challenging but not too ambitious&
% ma'e the scope of the research clear (will it refer to one company or be broader, for
e)ample*&
% clarify the meaning of any 'ey terms used&
% refer to practical or theoretical outcomes&
% be listed near the start of the project repory.
#ry to envisage the sort o conclusions which you might e&pect to arrive at. #hen ask yoursel:
% Are you li'ely to be able to get the evidence to justify these conclusions
% Are the conclusions worth the effort. +ut yourself in the position of a critic who says,
simply, ,"o what,.
*t the end o the pro!ect report" you should have a clear section e&plaining how you have
achieved the aims (or answered the 'uestions) laid out near the $eginning.
#eneral issues concerning research$ %hiloso%hy& etc
#he irst point to $e made is that the outcomes o a research pro!ect (the answers to the 'uestions
posed $y the researchers) may $e o a wide variety o dierent types. #he possi$ilities include:
% -niversal laws of the type which are common in natural science. (%g %Imc
5
. #.+
always improves proita$ility.) Such laws are very rare" or perhaps non(e&istent" in
management. #hey are not a realistic aim.
% "tatistical conclusions. (%g J;K o #.+ implentations ail. 1n average" on(the(!o$
training is more eective that class(room training.) #hese are common outcomes o
management research. -t is o$viously very important to speciy the scope o the research
(what training in which industries>) and the e&act meaning o key terms (on(the(!o$"
classroom" eective).
% .etailed analyses of particular situations (case studies*. %g a detailed case study o a
#.+ implementation which ailed might $e useul or understanding the causes o ailure
and so avoiding them elsewhere.
% Conceptual framewor's.
% /athematical and other models.
% 0ecommended procedures or methods.
)an you think o any other possi$ilities>
2nderstanding the %resent& %redicting the 0uture& and im%ro3ing the 0uture
0esearch pro!ects may seek to understand and e)plain the present and past situation" to predict
the future situation" or to recommend how to improve the future situation (sometimes called
2prescriptive2 conclusions)" or a com$ination o all three.
6or e&ample" consider a research pro!ect which aims to ind the $est 'uality strategy or a
particular company. #his might start $y developing an understanding o the e&isting pro$lems in
9
the company" and the eectiveness o the various possi$le 'uality strategies in use in the industry.
#his understanding may range rom a simple catalogue o pro$lems" to a deeper e&planation o
the sources o the pro$lems and the eectiveness o the various 'uality strategies.
#he ne&t step might $e to predict (roughly) the impact o the various possi$le strategies.
#hese predictions would $e $ased on the understanding o the e&isting pro$lems and the
eectiveness o the various possi$le strategies.
#hese predictions can then $e used to decide which strategy is likely to $e $est in the
sense that it will improve the companyHs perormance more than the others. #he research is aimed
at understanding" prediction and improvement" $ut improvement is" o course" the main goal.
Unortunately" most discussions o research methods in management are $ased airly
closely on similar discussions a$out the natural and social sciences ( which aim to understand and
predict" $ut not to improve. #his means that the aim o making improvements tends to $e ignored
in philosophical discussions. Ulrich (17L<" p. 19) claims that 2there is no ade'uate philosophical
$asis2 or this type o research. #his is serious $ecause the logical $asis o recommending
improvements is very dierent rom the logical $asis o understanding or predicting.
#here are two important dierences. #he irst is that i a change is made" the new
situation will $e dierent rom the e&isting situation" and so diicult to research directly. -t is
diicult to study the impact o a new idea which has not $een triedE #here are a num$er o ways
round this diiculty: the use o e&periments" action research and modelling (see $elow)" and
studying (or e&ample) other organisations which have tried the new idea. (#his is not possi$le" o
course" i the idea is really new.)
#he second point a$out making recommendations a$out what an organisation ought to
do to improve perormance is that this o$viously presupposes some value judgements. #hese are
2su$!ective estimates o worth2 (the +oc'et 1)ford .ictionary" )larendon Press" 177J): ie
assertions a$out how things are valued" or a$out what is good and what is $ad" and a$out which
goals an organisation or individual should strive or. ?ierent groups in an organisation" or
dierent stakeholders" may" o course" arrive at dierent value !udgements" and dierent
recommendations a$out what should $e done.
-t is important to try to $e as e&plicit as possi$le a$out the $asis o these value !udgments.
Chere the value !udgments depend on several dierent criteria" it may $e helpul to indicate how
each 'uality strategy (or whatever) scores against each criterion $y means o an 2options $y
criteria matri&2. (See also 0o$son" 177<" chapter 8 on 2%valuations2" and Geeney" 1775.)
Both o these points ( the act that the research has to study hypothetical situations" and
has to $e $ased on value !udgments ( mean that research which seeks to improve situations its
uneasily into the crude idea o the scientiic method known as positivism ( to which we turn ne&t.
(?espite this" 2management science2 is perhaps the main source o prescriptive management
researchE)
(ositi3ism and %henomenology& and similar distinctions
Positivism is the view that research should $e scientiic in a airly crude sense. #he reality
researched is viewed as e&ternal and o$!ective" and the methods used should $e 2value(ree2 and"
as ar as possi$le" 'uantitative. (#here are many dierent versions o positivism. #he conusion is
e&acer$ated $y the act that much o modern physics is ar closer to phenomenology than
positivism as it is usually understood" and some $ranches o management science have a lot to say
a$out values.)
Phenomenology 2stems rom the view that the world and HrealityH are not o$!ective and
e&terior" $ut that they are socially constructed and given meaning $y people2 (%aster$y(Smith et
al" 1771" page 5@" citing =usserl" 17@J). #his leads on to a style o research that involves detailed
interviews and other interactions with the actors involved in a situation" and appreciating" $ut not
J
necessarily predicting" the dierent perspectives and choices people adopt. -t typically involves an
in(depth study o a small sample o people which attempts to understand the e&perience o these
people 2rom the inside2 ( ie in terms o their su$!ective e&perience. #he researcher is inevita$ly
not independent o the situation under study" which may mean that dierent researchers come to
dierent conclusions. (?oes this matter>)
* phenomenological analysis is typically mainly 'ualitative in character rather than
'uantitative" and deterministic or statistical conclusions tend to $e shunned in avour o a
thorough analysis o a small num$er o cases ( which may" o course" illustrate possi$ilities which
could occur elsewhere. Positivistic research" on the other hand" typically involves larger samples"
which produce more relia$le statistical generalisations $ut at the cost o a shallower
understanding 2rom the outside2 ( ie in terms o e&ternally deined varia$les.
-t is not helpul to regard this as an either(or choice. *ny useul research is likely to draw
on both o$!ective acts and su$!ective e&periences" and to use $oth 'ualitative and 'uantitative
methods o analysis.
#here are other related concepts and distinctions ( hard and sot (0osenhead" 17L7)3" and
positivism and social constructionism (Burr" 1779" %aster$y(Smith et al" 5;;5). #he terms
M'uantitative2 and M'ualitativeN are oten used as um$rella terms or the two ends o the
spectrum.
#he meaning o many o these terms is rather haAy" so it is important to deine what you
mean when using them.
The degree o0 generality
%insteinHs amous e'uation %Imc
5
reers to all the matter in the universe at any time. -t is
perectly general.
*t the other e&treme the aim o devising the $est 'uality strategy or .0C ,td."
%msworth" %ngland in 1778 reers to one particular company at one particular time.
1$viously" other things $eing e'ual" the more general the research is the more useul it is.
=owever" other things rarely are e'ual. -n ields like management" general theories are oten too
vague to $e helpul in speciic situations" and they are also ar harder to set up. 6or this reason" it
is usually a good idea to ma'e your aims fairly specific 2 ie relating to one organisation or
sector or country. =owever it may $e worth adding a su$sidiary aim to generalise your
conclusions more widely (particularly i you are considering getting another !o$ or want to
pu$lish your indings).
Theories$ 4uilding& testing& amending and using
#he word theory means dierent things to dierent people. - think that anything which goes
$eyond a straight listing o the acts should $e counted as a theory. #his includes e&planatory
rameworks" generalisations" recommendations" mathematical models" etc. *ll useul research
involves theory in some orm ( there is a note on the meaning and role o theory in the appendi&.
Sometimes" the aim o the research is to develop theory rom scratch. #his is the
inductive approach: trying to derive generalisations and e&planations rom the data you collect. -n
its pure orm the researcher tries to orget any preconceptions and !ust let the data 2speak2. 4ou
will ind suggested tactics or this in $ooks on research methods" and in more detail in +iles and
=u$erman (177@).
#he other e&treme style o research involves starting with a theory" or hypothesis" and
then testing it. #he theory may come rom other researchers" or it may $e a hunch or a con!ecture.
#his is the hypothetico(deductive approach to research. Garl Popper is an inluential advocate o
this style o research (see Popper" 178L" or one o the many commentaries on PopperOs views).
-t is very important that the theory should $e very clearly deined. 6or e&ample 2Comen
8
are more intelligent than men2 could not $e properly tested without deining intelligence in
numerical terms" speciying which women the hypothesis reers to" and whether it reers to
average intelligence levels. Popper (178L) has stressed the importance o the hypotheses $eing
testa$le: he claims that the theories o +ar& and 6reud are useless $ecause their hypotheses
cannot $e tested.
-n practice" the $est approach is oten in the middle: a $it o induction" and a $it o testing
theory. #he result may $e an amended theory" or a theory adapted to a particular situation" or
conclusions a$out the value (or otherwise) o the theory in a particular conte&t.
Sometimes a research pro!ect will make use o a theory developed $y other researchers"
without trying to test or amend it. 6or e&ample" research into proita$ility and employee
empowerment might make use o measures o proita$ility and empowerment ( which are
themselves theories.
!he theories which play a part in your research are an important aspect of the project.
3ou should discuss these theories, and their role in your research, carefully.
(olitics and ethics
#he political issues surrounding access to data" and the impact o the results also need
considering. Cill you have access to the data you need> ?o you have to give guarantees o
conidentiality and i so does this matter> Chat i your conclusions are not to the liking o key
stakeholders>
Similarly" there are sometimes ethical dilemmas in research. #hese are o$vious in medical
research where" or e&ample" it is o$viously unair to withhold what is considered the $est
treatment in order to set up a controlled e&periment. -n management research" withholding
$eneits rom a comparison or control group may also $e considered unair. +ore generally"
e&cept in very special circumstances" it is considered unethical to mislead people involved in
research" to su$!ect them to stress" to invade their privacy" and so on. - interviewees are
promised they will not $e identiied in research reports" it is o$viously unethical to ail to do this.
Research design
=aving decided on the aims to $e achieved the ne&t stage is to design your research: in other
words devise a plan or achieving the aims. +uch o the most successul research uses a variety
o dierent methods. -t is $est to start without too many preconceptions concerning the $est
approach.
#here are three possi$le sources o inormation or research:
1 %mpirical sources: gathering inormation rom the real world. #his may $e primary data
that you have gathered yoursel" or secondary data gathered $y someone else ( eg
pu$lished statistics or company documents.
5 ,iterary sources: gathering inormation rom pu$lished $ooks and papers" and rom the
internet (see Stein" 1777).
< )onceptual analysis: analysing the meanings o concepts and their implications.
(+athematical analysis and model $uilding are conceptual in that they are concerned with
working out the detailed implications o assumptions.)
*lmost all pro!ects make some use o all three ( but the emphasis is usually on empirical
methods. =owever" your research report should always include a review o relevant research $y
others (the 2literature review2)3 and your research will also inevita$ly depend on a ramework o
concepts ( a 2conceptual ramework2) which should $e careully analysed and !ustiied. Chat do
you mean $y 2'uality2" 2competitive2 or whatever other terms are important or your research>
L
Em%irical methods
%mpirical research usually involves making choices in our areas:
1 *re you going to study the e&isting situation" or are you going to do an e&periment or a
2'uasi(e&periment2 ( ie change something and see what eect it has> %&periments and
'uasi e&periments are particularly useul or gathering support or recommendations.
5 Chat sort o sample are you going to take> ,arge sample" small sample or study o a
single case>
< *re you going to use a standard theory or ramework (and i so which>)" or are you
going to develop your own theory> -n either case" theories are important (see appendi&).
@ =ow are you going to gather the empirical data> #he possi$ilities include: written
'uestionnaires" interviews" o$servation" Mparticipant o$servationN" document and data
archive analysis" the internet" etc. )an you think o any others>
*ll o these choices deserve very careul consideration. ?onHt orget that you will pro$a$ly use
dierent approaches or dierent parts o your research.
#here are also some other possi$ilities which do not it neatly into this ramework (eg
computer simulation" role plays). !he important thing is to be fle)ible and use a variety of
methods to achieve your aims.
#he ollowing su$sections descri$e ive general patterns o research design: surveys"
e&periments and 'uasi(e&periments" case studies" action research" and modelling. #hese may
overlap ( a model may $e $uilt rom a case study or a survey" or an action research pro!ect may
make use o a survey ( and there are certainly other possi$ilities.
3ou should not be restricted by these4 good research generally uses a combination of
these patterns as well strategies which do not fit neatly into any of them.
Sur3eys
* survey involves the collection o inormation rom a (usually airly large) num$er o 2units2.
#hese units may $e people" or organisations" or towns" or amilies" or departments" etc3 the
inormation collected may $e o any kind ( eg inancial inormation or opinions in the case o
surveys o people" or inormation a$out num$ers o employees and organisational structures in
the case o a survey o organisations. * survey provides a snapshot o the situation as it is at a
particular time" usually with a view to analysing patterns and trends applying to the group as a
whole. +ost surveys are $ased on a sample o the population o interest (see notes on sampling
$elow). Surveys oten use 'uestionnaires to collect data" $ut interviews or o$servation may
sometimes $e preera$le. +any people seem to assume that an +asters degree pro!ect has to
include a 'uestionnaire survey $ut this is not so3 do not use a 'uestionnaire survey i it is not the
appropriate method or your purposes.
6urther reading in any $ook on research methods.
E5%eriments and !uasi6e5%eriments
Surveys provide a way o inding out a$out the present situation and what has happened in the
past. =owever" there are two ma!or diiculties with simply monitoring what is happening now
and what has happened in the recent past.
#he irst diiculty is that it may $e diicult to disentangle cause and eect. #here is
apparently (=u" 178<" p. L@) a strong and positive correlation $etween the num$er o $a$ies
$orn into amilies in =olland and ?enmark and the num$er o storksH nests on the roos o their
houses. ?oes this suggest that the storks are in act responsi$le or the $a$ies> 1$viously there is
a more plausi$le e&planation ( $ig amilies have $ig houses which provide more space or storks
to nest. =owever" you cannot make relia$le inerences a$out which actor is the underlying cause
7
rom the correlation o$served. #o test the hypothesis a$out storks increasing the num$er o
$a$ies" you would need to do an e&periment ( perhaps encouraging more storks to nest to see i
the num$er o $a$ies $orn increases. 4ou need to control the relevant varia$les (eg siAe o
houses" age and gender o occupants) so the comparison is a air one.
#he second diiculty is that things that have not happened cannot $e investigated. - the
$est solution is a com$ination o circumstances that have never arisen" no survey will ever ind it.
#he $est way round these diiculties is to design an e&periment. #his involves changing
something and then measuring the eect that this change has. #he simplest design or an
e&periment is the 2post(test only two group design2 (0o$son" 5;;5):
1 Set up an e&perimental and a comparison (control) group using random
assignment.
5 #he e&perimental groups gets the 2treatment23 the comparison group gets the
2comparison treatment2. -t is important to ensure that the two groups get roughly
the same amount o attention ( otherwise there is a possi$ility that any o$served
dierence may $e due to the M=awthorne eectN. #his is named ater a amous
e&periment in which it was discovered that any treatment ( including reversing a
previous treatment ( $rought improvements $ecause it indicated that the
e&perimenter was taking an interest in the people involved.
< Pive 2post(tests2 to see what the eect o the treatment is.
+ore comple& designs are o course possi$le (see 0o$son" 5;;5). #he random assignment is
important to reduce the likelihood o some actor other than the 2treatment2 $eing responsi$le or
any o$served improvement. #he results o an e&periment are then usually analysed $y means o a
statistical hypothesis test (see $elow).
%&periments are widely used in medicine" psychology" and to a lesser e&tent in education.
-n management" it is oten impossi$le to ollow a rigorous e&perimental design so quasi2
e)periments are oten used instead. .uasi(e&periments are deined $y )amp$ell and Stanley
(17J<) (cited $y 0o$son (5;;5)" p. 1<<) as
2a research design involving an e&perimental approach $ut where random assignment to
treatment and comparison groups has not $een used.2
6or e&ample" the success o an organisation using a particular type o 'uality management system
may $e compared with an organisation which does not use this type o system" or with the same
organisation $eore the system was implemented. -n either case the 2treatments2 ('uality system
or no 'uality system) were not allocated at random" so there is the (strong) possi$ility that some
other" uncontrolled" varia$le is responsi$le or any dierences ound. 6or this reason" 0o$son
(5;;5) does not recommend either o these designs" preerring more ela$orate designs (see
0o$son" 5;;5" pp. 1<J(1@J or details). #he important thing is to $e as sure as possi$le that the
lack o randomisation in 'uasi(e&periments is not likely to aect the validity o the results.
Case studies and small sam%le research
It often seems more useful to underta'e a detailed study of an individual case, or of a small
sample of cases, than to do a superficial study of a larger sample. #he cases may $e individual
people" organisations" neigh$ourhoods" pro!ects" events o various types" etc. -t is important to $e
clear a$out the purpose o the case study. -s it intended to $e typical o something more general"
or to $e a case o particular interest rom some speciic point o view>
-t is also important to make sure that your approach is systematic" and that you give
ade'uate attention to developing a suita$le conceptual ramework and list o research 'uestions.
)ase studies normally use multiple sources o evidence (eg interviews" o$servations" document
analysis" etc)" and should aim or a detailed (Min depthN) understanding o the chosen case(s).
6urther reading: 4in (177@).
1;
Action research
#raditional science seeks to keep the researcher separate rom the researched and their aims and
values in the interests o 2o$!ectivity2. *ction research is the name given to research which seeks
to integrate theory development and data collection with action in the sense o improving the
process $eing studied. #he action researcher would typically $e an active participant in the
process. #he o$vious danger here is that the particular interests o the researcher D actor will
encourage a $iased perspective: clearly you must try to reduce the likelihood o this happening.
(* counterargument to this starts rom the assertion that there is no such thing as an un$iased
perspective" !ust dierent $iases ....)
#here are dierent variants and interpretations o action research. 1ne simple possi$ility
would $e:
1 Study the e&isting situation
5 Plan how improvements could $e made.
< )arry out these improvements and analyse their eects and success. (#his step may $e a
'uasi e&periment.)
@ Study the new situation.
9 Po $ack to step 5" etc" etc.
-t is o$viously important to ensure that the researcherHs involvement in the process does not
compromise the validity o the results.
Modelling
+anagement science researchers oten seek to set up a model o" or e&ample" a stock control
system" or a series o cash lows" or a pro!ect. +odels are also important in many other areas
including" or e&ample" inance and 2soter2 disciplines such as marketing. =arding and ,ong
(177L) summarise @9 o these management models. +odelling is not treated as a standard type o
research in most te&ts on research methods ( you will need to consult $ooks such as Pidd (177J
or 5;;<).
Pidd (177J" p. 19) deines a model as
an e)ternal and e)plicit representation of part of reality as seen by people who wish to
use that model to understand, to change, to manage and to control that part of reality.
+odels may $e physical" mathematical or computer $ased. #hey are useul i e&perimenting
directly with reality is too diicult" costly or time(consuming. #hey are typically set up on the
$asis o empirical data and a 2common sense2 analysis o how the situation 2works2. +odels are
always simpler than reality: it is important to consider the appropriate degree o simpliication.
#he steps in a typical modelling pro!ect are:
(1) $uild the model3
(5) check its accuracy andDor useulness and ad!ust i necessary3
(<) use the model to understand" change" manage" control...
6urther reading: Pidd (177J) chapter 1.
A general design 0or a ty%ical Masters degree %ro.ect
+any ($ut $y no means all) pro!ects it the ollowing pattern:
Aim: #o ind a good strategy to 2improve2 F in org 4
/ethod
1 SurveyDcase studies o 1rg 4 to investigate pro$lems and opportunities
5 SurveyDcase studies to see how other organisations do F and which approaches work
well
< Based on (1)" (5)" the literature" and perhaps creative inspiration and consultations within
11
the organisation" devise a strategy likely to improve F
@ #ryDtestDpilotDmonitor the proposed strategy
7in-ing methods to research aims or !uestions
#o ensure that your methods are irmly linked to your research 'uestions (or aims)" it is a good
idea to draw a diagram which links each research 'uestion with the methods you plan to use to
answer it.
-n the diagrams $elow" the lines without arrows indicate the $reakdown o the research
aims. #he arrows indicate that the $o& at the start o the arrow is a means to help achieve the $o&
at the end o the arrow. #he arrows only indicate that a method will help with the aim or method
it points to" not that it will solve the pro$lem completely. #he dotted arrow is intended to signiy
that the help involved is likely to $e slight. (#his notation is due to Geeney" 1775).
5
#his diagram should help you to ensure that the methods you are proposing are likely to $e
suicient. #his is a matter o !udgement" o$viously. 4ou need to check each aim careully. -n this
e&ample" the lack o methods drawing on data rom 1rganisation 4 or assessing the
improvements rom the proposed strategy" and or devising and !ustiying the implementation
strategy" suggests that this plan is not ade'uate. 4ou are likely" or e&ample" to need some input
to the implementation strategy rom 1rganisation 4. #he ne&t diagram shows a possi$le
improvement.
15
<
Data collection methods
#here are many sources o data which you should consider ( see the section on %mpirical
methods a$ove). #his section contains very $rie notes on interviews and 'uestionnaires" and also
on sampling" which is important whatever you decide to do. 6or more detailed help" is essential
to consult a te&t$ook or other source o advice.
Inter3ie*s
#hese could play a part in surveys" or case studies" or e&periments" or action research. #hey
usually allow you ind out a$out the topic o interest in more depth than a 'uestionnaire" $ecause
people are likely to give more detail when talking than when writing" and it is possi$le to ask
'uestions to pro$e points o particular interest. -t is however necessary to $e organised: use a list
o 'uestions and prompts and decide how you are going to record the answers. #elephone and
group interviews are other possi$ilities to $ear in mind. *$ove all" remem$er that the idea is to
get a deep understanding o the issues in 'uestion.
B Crite a plan or schedule or the interviews" $ut treat it le&i$ly and $e prepared to modiy
it i appropriate. Chat are you going to ask and how> ?onHt orget that interviews are
particularly useul or open(ended 'uestions.
B 4ou will pro$a$ly want to pro$e some responses or more detail. Some such pro$es can
$e in the interview plan" $ut o$viously as you do not know what the interviewees will say"
you cannot cover all eventualities.
B -t is a good idea to record the interview so that you can 'uote interesting $its in the write(
up. 4ou must ask interviewees or their permission" o course. - a recording is not
possi$le" you will o$viously need to make very detailed notes.
B ?onHt orget to think a$out putting interviewees at ease.
B Cith interviews there may $e a danger that the interviewer inluences the interviewee.
?oes this matter and what can you do a$out it>
8uestionnaires
1<
+any $ooks and articles give advice on 'uestionnaires: you should consult one at an early stage
because designing good questionnaires is far more difficult than it may loo'. -t is essential to
test the 'uestionnaire and the proposed method o analysis $y means o a pilot survey $eore the
inal 'uestionnaires are sent out.
When designing questionnaires consider4
B %&actly what do you want to ind out>
B Chy should people ill it in> (*nonymity" conidentiality> 0eward or returning it>)
B Cill they tell the truth>
B ,ength and se'uence o 'uestions
B Cording: avoid leading" long" complicated 'uestions asking several things"
incomprehensi$le" unanswera$le" silly" rude" annoying 'uestions....
B #he covering letter e&plaining who you are and what the research is or.
!here are three main types of questions you can as' in a questionnaire4
B )losed 'uestions asking or a category. (Chich department are you in> ( tick the
appropriate $o&.) Be careul to ensure you have thought o all the categories3 you should
usually have a $o& at the end or 1ther 2 please specify.
B )losed 'uestions asking or a num$er. (=ow old are you> .uestions asking respondents
to rate their agreement with a series o statements on a 1 to 8 scale.)
B 1pen ended 'uestions. (Chat do you think o Q >) #he responses may either $e coded
or analysis (in which case it may $e $etter to use a closed 'uestion in the irst place)" or
simply read and used or 'uotations and as a means o coming to understand the
respondents.)
Particularly with closed 'uestions you need to cater or respondents who do not know the
answer. 4ou donHt want to orce them to make up an answerE
0emember that designing a good questionnaire is much more difficult than it loo's.
Common +roblems with questionnaires4
B ,ow response rate (Chat should you do a$out this>)
B #oo much inormation to analyse
B -nconclusive answers
B 4ou only ind out what people want to (and can) tell you
5inally, as' yourself, are you sure you need a questionnaire Would you fill it in yourself If not,
why not thin' again
Sam%ling
Ce oten talk a$out analysing data (igures" etc) and drawing graphs o data as i we were
interested in the data or its own sake. Usually this is not the case. Usually we are interested in
our data $ecause o what it tells us a$out a wider situation. So" or e&ample" an opinion poll
might ask 1;;; voters how they are going to vote in the ne&t election: the assumption $eing" o
course" that the voting pattern o the electorate as a whole will $e similar.
#he irst step is to decide e&actly where our interests lie. +opulation or universe are
terms used $y statisticians or the group comprising all the instances in which we are interested. -t
is important to $e very clear a$out the e&act nature o the population. 6or e&ample:
B employees in an organisation
B employees in all similar organisations
1@
B all the transactions which may $e carried out $y a sotware system (now and in the uture)
- the population is large or ininite we will need to use a sample: ie a su$set chosen as ar
possi$le to $e representative o the population as a whole. It is important in all investigations,
quantitative and qualitative, large scale and small scale, to be careful about the choice of a
sample.
%ven when it is apparently possi$le to look at every mem$er o the population ( ie to
carry out a census" the $eneits may not $e real. -n one survey o applications o 21;;K
inspection2 (1akland" 17LJ" p 9;)" 18K o deects on P)BHs were missed" and 59K on chest F(
rays (where a deect may represent a case o #B). #he pro$lems in each case were that the
necessity to check everything meant that the !o$ was done 'uickly and carelessly. -t is oten a
good idea to take a airly small sample and investigate this careully.
-n addition" populations are oten slightly wider than is apparent at irst sight. Ce might"
or e&ample" consider all the transactions perormed $y a computer system in the past week as
our population3 however a more useul perspective might $e to think o these transactions as a
sample o the possi$le transactions or which the system is designed. #his raises the 'uestion o
whether the past weekHs perormance is likely to typical or representative.
6rom the point o view o ensuring representativeness" two pro$lems may arise in
sampling:
1 #he method o selecting the sample may lead to an inevita$le bias (even with large
samples). -t is oten surprisingly diicult to o$tain an un$iased sample.
5 %ven i the method o selection does not lead to $ias" inevita$le random variations may
mean that the particular sample chosen is unrepresentative in some way. #his is known as
sampling error" and its siAe can $e analysed $y statistical methods: eg the <K error oten
'uoted or surveys o electorsH voting intentions with samples o around 1;;; is $ased on
a 79K statistical conidence interval (Cood" 5;;<). )onversely" the theory can $e turned
round to tell you how large a sample is necessary or a given degree o accuracy.
+ethods o sampling can $e divided into probability sampling (where the idea is to try to ensure
that the sample is representative $y controlling the pro$a$ility o each individual $eing chosen)"
and non2probability sampling (which does not use this principle). 6our important methods o
sampling are:
+robability sampling4
1 0andom sampling: sample chosen so that every mem$er o the population has an e'ual
chance o $eing selected" and every mem$er o the sample is chosen independently o
every other mem$er. -t also means that the sample is chosen without allowing the
investigatorHs (possi$ly su$conscious) preerences to inluence the choice. !his is the
standard on which most statistical theory is based. #o produce a random sample it is
necessary to have a num$ered list o the population ( this list is known as a sampling
frame. #hen the sample is chosen $y drawing random num$ers (see $elow) and selecting
the corresponding mem$ers o the population as the sample.
5 "tratified sampling: population divided into 2strata2 and a random sample o appropriate
siAe taken rom each o the strata. - done properly this should yield a slightly lower
sampling error $ut the dierence is oten very small. -t is generally only worth doing i it
easy to do or you want to compare results $y stratum. 4ou should also $ear in mind that
your sample will suer i you only take a ew o the strata. 6or e&ample" i you $ase a
sample o workers on !ust three companies" this sample will o$viously not encompass as
much variety as it would i you took a wider sample o companies.
19
6on2probability sampling4
< +urposive sampling: the researcherHs !udgment is used to choose individuals which are
thought to $e typical or o special interest. -t is oten a good idea to choose small samples
(eg or case studies) in this way3 or larger samples" the random or stratiied methods are
likely to produce more representative results.
@ 1pportunity or convenience sampling: taking the sample that you can get. #his is
eectively working $ackwards: the pro$lem then is deciding on the population to which
the results can $e generalised.
As a general principle random sampling is best for large samples (say 789*, whereas purposive
sampling is suitable for small samples. 0emember that the final sample may be smaller than
you anticipate because of non2return of questionnaires, etc.
0andom numbers (produced by a spreadsheet*
59J7 711@ 8;87 55;7 <LJ8 ;87< J788 <85; 191; 58J9 @;8@ 9L8L 9897 5<18 @989
J55@ 1<77 81J1 J7;< J@1@ 1875 979J 19@< <158 @L79 5LJ1 J81@ ;J8J ;J<9 8<77
<@5; 8L58 511J 8J85 198< ;J<5 997@ 11@7 7<5; 55LL 8J<@ J@J@ 5<8L J897 78<L
18<@ 1;J< 5L@L J@L7 L89; 11L7 9@7; 8L5J ;L1L 717J 9L9L @9LJ @875 15J; J955
1;<7 77<; 8781 5;75 L;8J 9JLJ L911 597L LJL8 8@87 7@<J ;J77 L5J@ 18<9 J9<5
;LJ; J<1< J1<5 8;;9 8;@9 11L< 91L< J@85 L;51 981J 8555 888< 9LLJ 8@8< <;<<
L7;; 5<L@ L599 7;1@ 15;7 LL78 5L5L 1@J1 8<77 ;J5< L758 <8L7 5;<; 177< 1;7@
858@ <99@ 5@<7 @<J; 17;;
Trust*orthiness or credi4ility
Chat makes research trustworthy> Chy should you $elieve or accept the conclusions> #he
concepts o validity, reliability, objectivity, triangulation and statistical hypothesis testing are all
relevant to this issue. #he most general concept is validity.
*s well as $eing trustworthy" research should" o course" also $e relevant and useful.
0eaders should not $e let asking 2So what>2.
9alidity
/alidity reers to the e&tent to which the results are valid ( ie true or well grounded. Pill
and Rohnson (1771" p 1J1) distinguish three types o validity:
1. -nternal validity is the e&tent to which the conclusions regarding cause and eect are
warranted.
5. Population validity is the e&tent to which conclusions can $e generalised to other people"
or other organisations" or other sampling units. #his is a matter o ensuring that samples
are likely to $e representative (see the notes a$ove).
<. %cological validity is the e&tent to which conclusions might $e generalised to social
conte&ts other than those in which data has $een collected.
#here is also ...
@. #he e&tent to which operational deinitions or indicators (eg deect rates as a deinition o
'uality3 -. tests as a measure o intelligence) relect the concept they are trying to
capture.
1J
Relia4ility
#his reers to the consistency o the research method. 6or e&ample would you get the same
answer i you repeated the research with a dierent sample" at a dierent time" or with dierent
o$servers or !udges> Suppose" or e&ample" your research involves coding responses to an open(
ended 'uestion on a 'uestionnaire. 4ou should check a sample o codes $y $ringing in a second
researcher. 4ou could then indicate the relia$ility o the coding scheme $y saying that the two
coders agreed on the code given to 79K (or whatever) o responses. #his provides the reader
with a simple assessment o how relia$le this aspect o the research is.
O4.ecti3ity
#his term reers to the e&tent to which research relects the reality o the 2o$!ects2 (including
people) under study" as opposed to the su$!ective e&perience o the researchers or o$servers. -n
practice" the method or checking whether an o$servation or assessment is o$!ective is to see i
dierent o$servers agree: i they do it is o$!ective" i they do not it is su$!ective in the sense that it
depends on the su$!ectivity o particular people. Physical measurements like weight or time are
o$!ective $ecause dierent o$servers will agree readily" whereas assessments o the 'uality o a
meal are more likely to $e su$!ective.
Some would say o$!ectivity is essential3 other would say that it is meaningless or
impossi$le in many conte&ts. ?o you think it is sensi$le to talk a$out the o$!ective 'uality o a
meal> 1n the other hand i you are interested in the amount o scrap produced" it seems sensi$le
to get as o$!ective a measure as possi$le.
Triangulation
)hecking your conclusions $y other methods. 6or e&ample" i 'uestionnaire results suggests that
particular managers are not motivated $y money" this could $e checked $y interviewing the
managers" and $y o$serving their $ehaviour (or records o their $ehaviour) when oered inancial
incentives.
Statistical hy%othesis tests
#hese provide a way o deciding i the evidence is strong enough. %&amples are the 2t test2"
analysis o variance (*S1/*) and the 2)hi s'uare test2. #hese tests are mathematically comple&"
and are very re'uently misunderstood and misinterpreted. ?espite this they are useul and widely
used. "tatistically significant means that the data cannot reasonably be attributed to chance
alone (ie to the accident of the particular sample chosen*. * signiicant result signifies a real
effect (and not !ust a sampling accident). #he signiicance level tells us how strong the evidence is
( with the lower levels indicating stronger evidence.
6or e&ample" the results $elow (+cPoldrick T Preenland" 1775) come rom a survey on
the service oered $y $anks and $uilding societies:
18
Aspect of service Ban's: mean
rating
Building "ociety:s
mean rating
;evel of significance
(p*
SympatheticDunder
standing
J.;@J J.<L7 ;.;;;
=elpul Driendly
sta
J.@79 J.78L ;.;;;
Sot too pushy J.<78 J.J@@ ;.;;<
#ime or decisions J.8<@ J.LJ9 ;.;5L
)onidentiality o
details
8.L<@ 8.88L SS
Branch manager
availa$le
9.75L J.;78 ;.;7;
#he data was o$tained rom a sample o customers who rated each institution on a scale ranging
rom 1 (very $ad) to 7 (very good.). #he a$ove si& dimensions are a selection rom the 55
reported in the paper. #he evidence is strongest in relation to the irst two varia$les and weakest
in relation to the least one. #he p values in the inal column o the ta$le give the estimated
pro$a$ility o o$taining the results which were actually o$served" or more e&treme ones" if there
is really no difference between ban's and building societies. (#here is a uller e&planation at
http:DDuserwe$.port.ac.ukDUwoodmDnmsDtest.doc .)
SS means not signiicant ( which in this ta$le means that the p value is greater than ;.1.
#he lower the p value the more convincing the evidence or a real dierence $etween $anks and
$uilding societies.D
-n many conte&ts (including the e&ample a$ove) Mconidence intervalsN provide an
alternative method o analysis ( which may $e more useul and user(riendly (Pardner and *ltman"
17LJ3 Cood" 5;;<).
Data analysis
#here are many methods o analysing data. 4ou should read up those that are appropriate to your
particular study.
*t one e&treme is statistical analysis. #he steps here are:
1 ?ecide what you are going to measure. )heck that the proposed measurements are valid
and sensi$le. - appropriate check the relia$ility o your measurements.
5 Produce diagrams andDor ta$les to show the values o your measurements and the
relationships and dierences $etween them. -t is more diicult than it might appear to
design ta$les and diagrams which are clear and unam$iguous ( ask someone else to
checkE
< - appropriate" do statistical hypothesis tests or work out conidence intervals to indicate
the likely eects o sampling error. (4ou may need help here.)
*t the other e&treme" the analysis o tapes o interviews" or open(ended 'uestions in
'uestionnaires" might simply consist o listening to the tapes" or reading the 'uestionnaire
responses" to try to understand the situation. #he report o the research would then include direct
'uotations (in MQN) rom the interviews" or the 'uestionnaires" as evidence or the assertions put
orward.
1L
#he weakness o this last style o research is that the particularly passages 'uoted may
give an unrepresentative impression. #he suspicion may $e that the researcher has chosen the
'uotes that conirm her (or his) pre!udices. )learly this type o analysis needs to $e $acked up $y
some urther evidence. -t is" however" a very useul method o providing a detailed analysis o
certain possi$ilities. 6or e&ample" a researcher investigating the use o a sotware package might
ind one individual using it in a particularly innovative manner: a detailed analysis o this one
instance may $e interesting $ecause it illustrates what is possi$le ( although it is in no sense
representative o the population as a whole.
#o use interview data" or data rom open(ended 'uestions on 'uestionnaires" to o$tain
more 'uantitative inormation a$out the re'uency with which phenomena occur" or the strength
o relationships" it is usually necessary to devise a coding scheme (see Saunders et al" 5;;<). #his
can $e used to give 'uantitative results on the percentage o individuals in each category" or the
num$er o times particular things are mentioned. #hese results can then $e analysed statistically
like any other 'uantitative results.
1ne issue to consider when analysing 2soter2 data rom interviews and participant
o$servation studies is the e&tent to which the conclusions should 2emerge2 rom the data without
the researcher imposing his or her preconceptions. #his is the grounded theory approach (see
Saunders et al" 5;;<3 0o$son" 5;;5). /arious methods have $een proposed or achieving this :
eg analytic induction (Saunders et al" 5;;<" <78(L3 0o$son" 5;;5" p. <55).
Chatever you do it is important to consider the validity and relia$ility (see a$ove) o your
inal conclusions.
6urther reading: +iles and =u$erman (177@).
Ty%es o0 measurement
/aria$les may $e numerical (eg salary)" ordinal (ie a rank ( eg +anchester UnitedHs position in
the league was 5nd)" or category varia$les (eg male or emale" make o car" etc). #ake care not to
manipulate results in ways that do not make sense. 6or e&ample there is little point in coding a
category varia$le (eg make o car) $y the num$ers 1" 5" <" @" etc and then taking the average ( it
wonHt mean anything.
Sumerical scales can $e urther su$divided into ratio and interval scales. 0atios make
sense in ratio scales $ut not interval scales. 6or e&ample it makes sense to say that one man earns
twice as much as another (earnings is a ratio scale)" $ut it does not make sense to say that a
temperature o 5; degrees )elsius is twice as hot a temperature o 1; degrees since temperature
is not a ratio scale3 the Aero point is ar$itrary ( the e'uivalent 6ahrenheit temperatures are 9; and
JL which are not in the same ratio.
Com%uter so0t*are
#he most useul type o package is a spreadsheet. %&cel is particularly good $ecause o the wide
range o statistical unctions and procedures which it incorporates. Put each record (individual
rom a sample) in a separate row with ield headings at the top. 6or e&ample:
S*+% S%F =%-P=# C%-P=#
Bill + 1.@8 1<5
Susan 6 1.71
+andy 6 1.@9 <L
*void the temptation to include ancy ormatting" to leave rows to improve spacing" etc. *ny rills
you include may cause pro$lems when you try to analyse your data.
17
- any data is missing (eg SusanHs weight) leave the cell $lank. ?o not enter ;. 4esDno is
$est coded as 1 or yes and ; or no3 then the average o the column will give you the proportion
answering yes.
4ou may $e a$le to do all your analysis with a spreadsheet. ?onHt orget that spreadsheets
will sort data. - you want to see how males dier rom emales" you can sort the data on this
ield. Spreadsheets are also good or working out averages" drawing $ar charts and other
diagrams" etc.
=owever" i the statistical analysis you need is at all comple&" it may $e worth transerring
the data to a statistical package such as SPSS (Statistical Package or the Social Sciences).
6urther reading: Cood (5;;<) contains $rie notes on the use o %&cel and SPSS or
analysing data.
Writing the re%ort
* standard layout is:
B *$stract
B *cknowledgments (i any)
B )ontents
B -ntroduction (including $ackground and conte&t ( this would normally lead on to the aims
in the ne&t chapter)
B *ims o the pro!ect (what you intend to achieve)
B ,iterature review (briefly and critically reviews relevant previous research and discusses
its relation to your study)
B 0esearch design or method (what you did and why)
B -nvestigation results and analysis (may $e split into several chapters)
B )onclusions and recommendations (possi$ly two chapters). 4ou should also discuss the
limitations o the research and possi$ly include suggestions or uture e&tensions.
B 0eerences (must ollow one o the standard ormats)
B *ppendices (supporting material to which readers may want to reer : eg 'uestionnaires"
e&amples o interview transcripts)
=owever" many pro!ects are not standard so you should eel ree to ad!ust this pattern i
appropriate.
Chatever the structure o your report" you should" as ar as possi$le" ensure that readers
can check your analysis to see i they accept your conclusions (put details in appendices). *$ove
all" please ensure that the report is clear" concise and does not e&ceed the permitted length.
-t is important to descri$e and discuss all important aspects o your empirical research:
details o 'uestionnaire surveys and interviews" sotware used" methods o analysis" and so on.
#he reader should $e a$le to ollow what you did" and how you derived your conclusions. #his
should ena$le the reader to decide how trustworthy your research is" and perhaps repeat it in
another conte&t. 0emem$er that i your research is well designed and competently carried out"
this should $e clear rom the report.
*ll $ooks and other sources should $e clearly reerenced using one o the standard styles.
#here is a lealet on this availa$le rom the li$rary" $ut you may ind it easier to copy the style
used in a particular academic paper. -n my view the easiest style is to reer to works in the te&t $y
the authorHs name and the date o pu$lication only ( or e&ample" Plato (179J) ( and then to list
the pu$lications in alpha$etical order o authorsH names at the end. %very reerence you give in the
te&t should appear in the list o reerences at the end ( check or Plato (179J) in the reerences at
the end o this document. (#he date is the date o the publication o the version to which you
reerred3 o$viously Plato did not write in 179J.) Sotice that $ooks and !ournal articles are mi&ed
up in this list o reerences3 otherwise you would not know which list Plato (179J) is in. Sote also
5;
the style o $ook and !ournal articles (eg #horpe and +oscarola" 1771) in this list o reerences.
The critical attitude
1ne o the distinguishing characteristics o good research is that as much as possi$le is su$!ected
to critical analysis. 4ou should 'uestion as much as possi$le. - the o$!ective o the pro!ect is to
derive a 2good2 strategy or a particular purpose" what does 2good2 mean> Cho says and how do
you know> Chy is this method appropriate> Chat are the potential laws with this method and
how did you try to overcome them> Chat are the main weaknesses o your research" and other
research in the ield> !ry and anticipate and answer all possible criticisms of your research.
(u4lishing your research
- you think your pro!ect deserves a wider audience you should consider pu$lishing it in a !ournal
or in some other ormat. *sk your supervisor or advice.
Chec-lists *hen starting a %ro.ect … and 0inishing it
!hese are my suggestions for chec'ing your initial project proposal:
1 Chat outputs do you e&pect> Crite down some e&amples o the sort o conclusions and
recommendations you might e&pect at the end o your pro!ect.
5 So what> -s the world ( or at least part o it ( going to $e a $etter place once these
conclusions and recommendations have $een reached>
< *re you likely to $e a$le to get the right data" and enough data" to !ustiy these
conclusions. Chat i a key stakeholder doesnHt like your results" conclusions or
recommendations> =ave you access to all the inormation you need> Cill the inormation
$e suiciently accurate and relia$le>
@ *re the aims challenging $ut not so am$itious as to $e impossi$le with the limited
resources (time" etc) at your disposal> -t is oten a good idea to have a airly restricted
ocus that is analysed in depth.
9 *re your research methods appropriate to achieve the aims> - you have" say" three aims"
you must make sure that you have considered the methods or achieving all three o them.
And at the end of the project you should chec' that:
1 4our research aims" literature" analysis and conclusions are clearly linked together. -t is
important to $e very clear a$out how your conclusions and recommendations ollow rom
your analysis" and achieve the aims you set yoursel at the start.
5 0emem$er that you are reporting a research pro!ect. -t should $e clear rom your write(
up that you have done some useul" systematic and rigorous research. +ake sure that you
give enough detail or this to $e clear.
< 6inally" check that your written pro!ect satisies the re'uirements in the guidelines you
have $een given.
Re0erences
<eneral te)ts on research methods include "aunders et al (8==>*, 0obson (8==8*, ?asterby2
"mith et al (8==8*.
Burr" /. (1779). *n introduction to social constructionism. ,ondon: 0outledge.
%aster$y(Smith" +." #horpe" 0." T ,owe" *. (5;;5). +anagement 0esearch: an introduction
(5
nd
edition). ,ondon: Sage.
%aster$y(Smith" +." #horpe" 0." T ,owe" *. (1771). +anagement 0esearch: an introduction.
51
,ondon: Sage.
6eyera$end" P. G. (1789). *gainst method: an outline o an anarchistic theory o knowledge.
,ondon: Sew ,et Books.
Pardner" +. R." T *ltman" ?. P. (17LJ). )onidence intervals rather than P values: estimation
rather than hypothesis testing. British +edical Rournal" 575" 8@J(89;.
Pill" R." T Rohnson" P. (1771). 0esearch +ethods or +anagers. ,ondon: Paul )hapman
Pu$lishing ,td.
=arding" S." T ,ong" #. (177L). +B* management models. *ldershot: Power.
=u" ?. (178<). =ow to lie with statistics. Penguin.
=usserl" %. (17@J). Phenomenology in %ncyclopaedia Britannica" 1@th edition" /ol 18"
J77(8;5.
Geeney" 0. ,. (1775). /alue(ocused thinking: a path to creative decisionmaking. )am$ridge"
+assachusetts: =arvard University Press.
+cPoldrick" P. +." T Preenland" S. R. (1775). )ompetition $etween $anks and $uilding
societies. British Rournal o +anagement" <" 1J7(185.
+iles" +. B." T =u$erman" *. +. (177@). .ualitative data analysis (5nd edition). ,ondon:
Sage.
1akland" R. S. (17LJ). Statistical process control. ,ondon: =einemann.
1akland" R (17L7). #otal .uality +anagement. 1&ord" =einemann Proessional.
Pidd" +. (177J). #ool or thinking: modelling in management science. )hichester: Ciley.
Pidd" +. (5;;<). #ool or thinking: modelling in management science (5
nd
edition). )hichester:
Ciley.
Plato (179J). +eno (trans: Puthrie" C G )). =armondsworth: Penguin.
Popper" G. (178L). )on!ectures and 0eutations. ,ondon: 0.G.P.
.uinn" R B3 +intA$erg" =3 Rames" 0 + (17LL). #he strategy process: concepts" conte&ts and
cases. Prentice =all.
0o$son" ). (5;;5). 0eal Corld 0esearch (5
nd
edition). 1&ord: Blackwell.
0osenhead" R. (17L7). 0ational analysis or a pro$lematic world: pro$lem structuring methods
or uncertainty" comple&ity and conlict. )hichester: Ciley.
0ussell" Bertrand (17J1). =istory o Cestern Philosophy. Peorge *llen T Unwin.
Saunders" +." ,ewis" P." T #hornhill" *. (5;;<). 0esearch methods or $usiness students (<
rd


edition). =arlow: Pearson %ducation.
Stein" S. ?. (1777). ,earning" teaching and researching on the internet: a practical guide or
social scientists. =arlow: *ddison Cesley ,ongman.
#horpe" 0." T +oscarola" R. (1771). ?etecting your research strategy. +anagement %ducation
and ?evelopment" 55(5)" 158(1<<.
Ulrich" C. (17L<). )ritical heuristics o social planning. Bern and Stuttgart: =aupt.
Cood" +. (5;;<). +aking sense o statistics: a non(mathematical approach. Basingstoke:
Palgrave.
4in" 0. (177@). )ase study research: design and methods (5nd edition). #housand 1aks" )*:
Sage.
55
A((ENDICES
A note on :theory:
* MtheoryN is deined $y the Concise 1)ford .ictionary as a Msupposition or system o ideas
e&plaining something" esp. one $ased on general principles independent o the particular thing to
$e e&plained.N #his clearly hinges on the meaning o Me&plainN ( which is deined as Mmake
intelligi$leN.
*ccording to 0ussell (17J1" p. 95)" the word theory is derived rom an 1rphic word
which can $e translated as Mpassionate sympathetic contemplationN3 at irst sight this is very
dierent rom the modern meaning $ut in act it its well with the ethos o" or e&ample" the
research method o participant o$servation.
#heory is oten contrasted with MactsN and what happens Min practiceN. * act is Ma thing
that is known to have occurred" to e&ist or $e trueN" and Min practiceN means Mwhen actually
applied" in realityN. * theory is thus a system o ideas which e)plains something" or makes it
intelligible" whereas acts and practice are simply the reality o what happens. (=owever" the
physicist" Sir *rthur %ddington" dismisses the common assumption that acts are more certain
than theory in physical science: 24ou should never $elieve any e&periment VactW until it is
conirmed $y theory2 ( 'uoted in #he <uardian" Ranuary 8" 177<).
#o give a concrete e&ample" it might $e a act that a irmHs sales have increased $y a
particular amount. * theory to e&plain this might $e the assertion that the increase in sales is the
result o improved 'uality in the products sold. #he system o ideas which orms this theory is the
act that 'uality levels have improved" and the assertion that" in these circumstances" improved
'uality is likely to lead to increased sales. #he theory is useul $ecause it gives us a means o
predicting when sales are likely to rise and so o increasing sales in new situations. * list o acts
and o what happens in practice may $e interesting3 however to predict and control in new
situations" theory is needed. #his reason or going $eyond acts and a simple description o
practice" to theory" seems" to me" unanswera$le.
*ccording to .uinn" +intA$erg and Rames (17LL) 2theories are useul $ecause they
shortcut the need to store masses o data ... it is easier to remem$er a simple ramework ... than
to remem$er every detail you ever o$served2 (p. &viii). =owever" this misses the most important
unction o theory which is to help cope with new situations which you have not yet o$served.
=owever" even apart rom this reason or using theory as a means o going $eyond the
given acts" theory is necessary or deining the MactsN. #he a$ove e&ample depends on a way o
measuring 'uality. #his can $e done in various ways ( $y reported deect rates" $y customer
satisaction" or $y some other means. 1$viously" we need a system o ideas deining 'uality
$eore we can even claim to detect an increase. #he re'uired theory might $e ormal academic
theory" or it might $e provided $y Mcommon senseN. But in either case it is still a theory. #he same
is true o many other MactsN: proita$ility can only $e deined $y reerence to theories o
accounting" acts a$out organisational structures can only $e deined $y reerence to the
appropriate theories. %ven a simple 'uestionnaire designed to elicit an attitude or an opinion
depends on the theory that people give true (or valid or meaningul) answers to such 'uestions.
(#his is oten a rather du$ious theory.) -n all these cases the acts are deined $y the underlying
theory. #he acts cannot even e&ist without the theory" and dierent theories are likely to give rise
to dierent acts. Chether this applies to all acts" or !ust some acts" is an issue which need not
concern us here. #he important thing is that it applies to many acts o interest to management
researchers.
#his means that the use o theory is inevita$le and it is clearly important to use the $est
theory or the purpose in hand.
5<
Ty%es and le3els o0 theory
Part o the diiculty in discussing theory is that the single term encompasses a very $road range.
%&amples o theories are the simple assertion that an improvement in 'uality led to an increase in
sales (see a$ove)" theories a$out how 'uality can $e measured and monitored" mathematically
$ased theories such as the model or calculating the economic order 'uantity" the theory that
speciying o$!ectives clearly increases the chances o a pro!ect succeeding" the theory that there
are particular categories o organisation" and" on a much more am$itious scale" the theory o total
'uality management (1akland 17L7). #hese are all theories in the sense a$ove. #hey are all useul
or deining the acts and or providing e&planations a$out" or e&ample" what to do in given
situations.
#heories may dier in their source: some come rom academic pu$lications" while others
may $e derived rom common sense. #hey dier in their level o generality. #hey dier in the
sense in which they Me&plainN things: sometimes the e&planation leads to a prediction (ollowing
the #.+ way will lead to improvements in 'uality which will lead to increases in sales)3
sometimes it merely categorises the possi$ilities ( which is an essential prere'uisite or
understanding and managing a situation. #heories may $e stated in ormal mathematical terms or
in inormal terms" which allow or even encourage diering interpretations. #heories dier in many
other ways. But they are all theories.
#he pro$lem or the researcher is that o choosing" creating" or adapting" the $est theory
or the purpose in hand. -t is important to investigate all the possi$ilities and make the selection
careully.
Theories may 4e *rong or inade!uate
Scientists tend to think o the current theory as the MtruthN. =owever" even the history o physical
science indicates that this is likely to $e a very limited perspective: there are many old MtruthsN (
the earth $eing the centre o the universe" atoms $eing unsplitta$le" matter indestructi$le ( which
have $een replaced $y contradictory new MtruthsN. -n management" ew" i any" theories command
respect rom everyone. #heories o management are much more o$viously alli$le and or this
reason should not $e taken too seriously.
Conclusions
Chat is the relationship $etween theory and management research> - think that the discussion
a$ove demonstrates that:
1 #heories are necessary as a $ackground or a research pro!ect to deine the concepts and
terms in which the research is phrased. ?enying this does not make it less true3 it !ust
means that the implicit theories underlying the research will $e unacknowledged"
uncriticised" and" very likely" 'uite unsuita$le or the !o$.
5 #he only useul aim or research is to make a contri$ution to theory" since a simple list o
acts or practices is o little use. #he ollowing seem to me to $e the possi$le types o
contri$ution:
(a) ?emonstrating that an e&isting theory applies to a particular situation and
showing how it can $e used in this situation: or e&ample an application o #.+
theory F to 1rganisation 4.
($) +odiying" ela$orating or e&tending an e&isting theory: or e&ample
demonstrating that #.+ theory F" when applied to organisations o type 4"
needs modiying in a particular way.
(c) )reating a new theory.
(d) ?emonstrating that an e&isting theory is wrong or useless.
5@
(#he reader should $ear in mind that the theory presented here" a$out the role o theory in
management research" is as alli$le as any other theory and should $e not accepted uncritically. -t
represents my analysis3 others may disagree.)
An e5am%le to sho* the analysis o0 !uestionnaire data
#he 'uestionnaire was to o$tain eed$ack rom students on a course. -t comprised one 'uestion
asking or the studentHs tutorial group (PP ( an 2independent varia$le2)" 51 'uestions asking or
ratings o dierent aspects o the course on a 1(8 scale (the 2dependent varia$les2)" and two open
ended 'uestions which were analysed separately. #he data was entered in a spreadsheet" and then
the analysis was carried out using SPSS (Statistical Package or the Social Sciences). 0eno was
a reerence num$er written on each 'uestionnaire to identiy it.
SPSS was used to produce histograms" means and standard errors or each o the 51
'uestions" and a $reakdown o the scores $y tutorial group and an analysis o variance to assess
the signiicance o these results. -t could also give other statistics such as standard deviation"
skewness" kurtosis" minimum" ma&imum" etc. #here were a total o JJ pages o output o which
one is $elow. (* lengthier 'uestionnaire or a more detailed analysis can easily result in hundreds
o pages o output.)
-t would also $e possi$le to use a spreadsheet to do some" i not all" o the analysis.
59
To% le0t o0 data s%readsheet
0eno PP .1 .5 .< .@ .9 .J .8 .L
1 11 1 @ @ @ 5 < 1
5 5 1 < 1 @ @ @ @
< 9 @ @ @ < 9 <
@ @ @ 9 9 J J 9 9
9 < 5 @ 5 @ < 5
(Sote that missing data is indicated $y leaving the cell $lank.)
Analysis o0 9ariance
Sum o0 Mean ; ;
Source D/;/ S!uares S!uares Ratio (ro4/
Between Proups 1; @7.<8L7 @.7<87 5.@;99 ;.;558
Cithin Proups @< LL.5J7< 5.;95L
#otal 9< 1<8.J@L1
#rou% Count Mean ,+ (ct Con0 Int 0or Mean Minimum Ma5imum
Prp 1 < J.<<< @.L771 #o 8.8J8J J.;;;; 8.;;;;
Prp 5 < @.<<< 5.L771 #o 9.8J8J @.;;;; 9.;;;;
Prp < 9 <.@;; .9@19 #o J.59L9 1.;;;; J.;;;;
Prp @ < @.<<< (.L<87 #o 7.9;@9 5.;;;; J.;;;;
Prp 9 < <.;;; (1.7JL< #o 8.7JL< 1.;;;; 9.;;;;
Prp J 11 <.J<J 5.8855 #o @.9;;9 5.;;;; J.;;;;
Prp 8 L @.9;; <.9;;L #o 9.@775 5.;;;; J.;;;;
Prp L @ <.;;; .;7@7 #o 9.7;91 1.;;;; 9.;;;;
Prp 7 8 5.1@5 1.<1;8 #o 5.789; 1.;;;; <.;;;;
Prp1; @ <.9;; 1.7;LL #o 9.;715 5.;;;; @.;;;;
Prp11 < <.JJJ (.15L; #o 8.@J1< 5.;;;; 9.;;;;
#otal 9@ <.JL9 <.5@9< #o @.1591 1.;;;; 8.;;;;

Sponsor Documents

Or use your account on DocShare.tips

Hide

Forgot your password?

Or register your new account on DocShare.tips

Hide

Lost your password? Please enter your email address. You will receive a link to create a new password.

Back to log-in

Close