CHAPTER8SelectingaStudyDesign

Inthischapteryouwilllearnabout:

ThedifferencesbetweenquantitativeandqualitativestudydesignsCommonstudydesignsinquantitativeresearchandwhentousethemCommonstudydesigninqualitativeresearchandwhentousethemThestrengthsandweaknessesofdifferentstudydesigns

Keywords:actionresearch,after-onlydesign,before-and-afterstudydesign,blindstudies, case studies, cohort studies, control studies, cross-sectional study design,double-blind studies, experimental study design, feminist research, focus studies,longitudinal studies, non-experimental studies, panel studies, prospective studydesign, quasi-experimental studies, reflective journal, retrospective studies, semi-experimentalstudies,trendstudies.

Differencesbetweenquantitativeandqualitativestudydesigns

Inthischapterwewilldiscusssomeofthemostcommonlyusedstudydesignsinbothquantitativeandqualitative research. Overall, there are many more study designs in quantitative research than inqualitativeresearch.Quantitativestudydesignsarespecific,wellstructured,havebeentestedfortheirvalidity and reliability, and can be explicitly defined and recognised. Study designs in qualitativeresearcheitherdonothavetheseattributesorhavethemtoalesserdegree.Theyarelessspecificandprecise,anddonothavethesamestructuraldepth.Differencesinphilosophicalperspectivesineachparadigmcombinedwiththeaimsofastudy,toa

largeextent,determinethefocus,approachandmodeofenquirywhich,inturn,determinethestructuralaspects of a studydesign. Themain focus in qualitative research is to understand, explain, explore,discover and clarify situations, feelings, perceptions, attitudes, values, beliefs and experiences of agroupofpeople.Thestudydesignsarethereforeoftenbasedondeductiveratherthaninductivelogic,are flexible and emergent in nature, and are often non-linear and non-sequential in theiroperationalisation.Thestudydesignsmainlyentailtheselectionofpeoplefromwhomtheinformation,throughanopenframeofenquiry,isexploredandgathered.Theparametersofthescopeofastudy,and

informationgatheringmethodsandprocesses,areoftenflexibleandevolving;hence,mostqualitativedesigns are not as structured and sequential as quantitative ones. On the other hand, in quantitativeresearch, themeasurementandclassificationrequirementsof the information that isgathereddemandthatstudydesignsaremorestructured,rigid,fixedandpredeterminedintheirusetoensureaccuracyinmeasurementandclassification.Inqualitativestudiesthedistinctionbetweenstudydesignsandmethodsofdatacollectionisfarless

clear.Quantitativestudydesignshavemoreclarityanddistinctionbetweendesignsandmethodsofdatacollection. In qualitative research there is an overlap between the two. Some designs are basicallymethodsofdatacollection.Forexample,in-depthinterviewingisadesignaswellasamethodofdatacollectionandsoareoralhistoryandparticipantobservation.One of themost distinguishing features of qualitative research is the adherence to the concept of

respondent concordance whereby you as a researcher make every effort to seek agreement of yourrespondents with your interpretation, presentation of the situations, experiences, perceptions andconclusions. In quantitative research respondent concordance does not occupy an important place.Sometimes it is assumed to be achieved by circulating or sharing the findings with those whoparticipatedinthestudy.The‘power-gap’betweentheresearcherandthestudypopulationinqualitativeresearchisfarsmaller

than in quantitative research because of the informality in structure and situation in which data iscollected.In quantitative research enough detail about a study design is provided for it to be replicated for

verificationandreassurance.Inqualitativeresearchlittleattentionispaidtostudydesignsortheotherstructuralaspectsofastudy,hencethereplicationofastudydesignbecomesalmost impossible.Thisleads to the inability of the designs to produce findings that can be replicated. Findings throughquantitativestudydesignscanbe replicatedandretestedwhereas thiscannotbeeasilydonebyusingqualitativestudydesigns.Another difference in the designs in qualitative and quantitative studies is the possibility of

introducing researcher bias. Because of flexibility and lack of control it is more difficult to checkresearcherbiasinqualitativestudies.Study designs in each paradigm are appropriate for finding different things. Study designs in

qualitativeresearcharemoreappropriateforexploringthevariationanddiversityinanyaspectofsociallife,whereasinquantitativeresearchtheyaremoresuitedtofindingouttheextentofthisvariationanddiversity. If your interest is in studying values, beliefs, understandings, perceptions, meanings, etc.,qualitativestudydesignsaremoreappropriateastheyprovideimmenseflexibility.Ontheotherhand,ifyour focus is tomeasure themagnitudeof thatvariation, ‘howmanypeoplehaveaparticularvalue,belief, etc.?’, the quantitative designs are more appropriate. For good quantitative research it isimportant thatyoucombinequantitativeskillswithqualitativeoneswhenascertaining thenatureandextentofdiversityandvariationinaphenomenon.Intheauthor’sopinion,thequalitative–quantitative–qualitativeapproachtoresearchiscomprehensiveandworthconsideration.Thisinvolvesstartingwithqualitative methods to determine the spread of diversity, using quantitative methods to quantify thespreadandthengoingbacktoqualitativetoexplaintheobservedpatterns.Asalreadystated,theauthordoes not recommend your locking yourself into either the qualitative or quantitative paradigm and,though you may have your preference, it is the purpose that should determine the choice betweenquantitative and qualitative study designs. If you already know (from previous studies or practiceknowledge) the nature of diversity in any area of interest to you, knowledge about its extent can bedeterminedonlybyusingquantitativemethods.Inmostcaseswhereyouwanttoexploreboth,youneedtousemethodsthatfallinthedomainofbothparadigms.

Studydesignsinquantitativeresearch

Someofthecommonlyuseddesignsinquantitativestudiescanbeclassifiedbyexaminingthemfromthreedifferentperspectives:

1. thenumberofcontactswiththestudypopulation;2. thereferenceperiodofthestudy;3. thenatureoftheinvestigation.

Every study design can be classified from each one of these perspectives. These perspectives arearbitrary bases of classification; hence, the terminology used to describe them is not universal.However, thenamesof thedesignswithineachclassificationbaseareuniversallyused.Note that thedesignswithineachcategoryaremutuallyexclusive; that is, ifaparticularstudy iscross-sectional innature it cannotbeat the same timeabefore-and-afteror a longitudinalstudy, but it canbeanon-experimental or experimental study, as well as a retrospective study or a prospective study. SeeFigure8.1.Anothersectionhasbeenadded to the threesections listedabove titled ‘Others–somecommonly

usedstudydesigns’.Thissectionincludessomecommonlyuseddesignswhicharebasedonacertainphilosophyormethodology,andwhichhaveacquiredtheirownnames.

Studydesignsbasedonthenumberofcontacts

Basedonthenumberofcontactswiththestudypopulation,designscanbeclassifiedintothreegroups:

1. cross-sectionalstudies;2. before-and-afterstudies;3. longitudinalstudies.

FIGURE8.1Typesofstudydesign

Thecross-sectionalstudydesign

Cross-sectionalstudies,alsoknownasone-shotorstatusstudies,arethemostcommonlyuseddesignin the social sciences. This design is best suited to studies aimed at finding out the prevalence of aphenomenon,situation,problem,attitudeorissue,bytakingacross-sectionofthepopulation.Theyareusefulinobtaininganoverall‘picture’asitstandsatthetimeofthestudy.Theyare‘designedtostudysomephenomenonbytakingacross-sectionofitatonetime’(Babbie1989:89).Suchstudiesarecross-sectionalwithregardtoboththestudypopulationandthetimeofinvestigation.Across-sectionalstudyisextremelysimpleindesign.Youdecidewhatyouwanttofindoutabout,

identifythestudypopulation,selectasample(ifyouneedto)andcontactyourrespondentstofindoutthe required information. For example, a cross-sectional designwould be themost appropriate for astudyofthefollowingtopics:

TheattitudeofthestudypopulationtowardsuraniummininginAustralia.Thesocioeconomic–demographiccharacteristicsofimmigrantsinWesternAustralia.TheincidenceofHIV-positivecasesinAustralia.Thereasonsforhomelessnessamongyoungpeople.

Thequalityassuranceofaserviceprovidedbyanorganisation.Theimpactofunemploymentonstreetcrime(thiscouldalsobeabefore-and-afterstudy).Therelationshipbetweenthehomeenvironmentandtheacademicperformanceofachildatschool.Theattitudeofthecommunitytowardsequityissues.Theextentofunemploymentinacity.Consumersatisfactionwithaproduct.Theeffectivenessofrandombreathtestinginpreventingroadaccidents(thiscouldalsobeabefore-and-afterstudy).Thehealthneedsofacommunity.Theattitudesofstudentstowardsthefacilitiesavailableintheirlibrary.

Asthesestudiesinvolveonlyonecontactwiththestudypopulation,theyarecomparativelycheaptoundertakeandeasytoanalyse.However,theirbiggestdisadvantageisthattheycannotmeasurechange.Tomeasurechangeitisnecessarytohaveatleasttwodatacollectionpoints–thatis,atleasttwocross-sectionalstudies,attwopointsintime,onthesamepopulation.

Thebefore-and-afterstudydesign

Themainadvantageofthebefore-and-afterdesign(alsoknownasthepre-test/post-testdesign)isthatitcanmeasurechange inasituation,phenomenon, issue,problemorattitude. It is themostappropriatedesign formeasuring the impact or effectiveness of a programme.A before-and-after design can bedescribedas two setsof cross-sectionaldata collectionpointson the samepopulation to findout thechange in the phenomenon or variable(s) between two points in time. The change is measured bycomparingthedifferenceinthephenomenonorvariable(s)beforeandaftertheintervention(seeFigure8.2).

FIGURE8.2Before-and-after(pre-test/post-test)studydesignAbefore-and-afterstudyiscarriedoutbyadoptingthesameprocessasacross-sectionalstudyexcept

that it comprises two cross-sectional data sets, the second being undertaken after a certain period.Depending upon how it is set up, a before-and-after study may be either an experiment or a non-experiment.Itisoneofthemostcommonlyuseddesignsinevaluationstudies.Thedifferencebetweenthe two sets of data collection pointswith respect to the dependent variable is considered to be theimpactoftheprogramme.Thefollowingareexamplesoftopicsthatcanbestudiedusingthisdesign:

Theimpactofadministrativerestructuringonthequalityofservicesprovidedbyanorganisation.Theeffectivenessofamarriagecounsellingservice.Theimpactofsexeducationonsexualbehaviouramongschoolchildren.Theeffectofadrugawarenessprogrammeontheknowledgeabout,anduseof,drugsamongyoungpeople.Theimpactofincentivesontheproductivityofemployeesinanorganisation.Theimpactofincreasedfundingonthequalityofteachinginuniversities.Theimpactofmaternalandchildhealthservicesontheinfantmortalityrate.Theeffectofrandombreathtestingonroadaccidents.Theeffectofanadvertisementonthesaleofaproduct.

Themainadvantageofbefore-and-afterdesignisitsabilitytomeasurechangeinaphenomenonortoassess the impact of an intervention. However, there can be disadvantages which may not occur,individuallyorcollectively,ineverystudy.Theprevalenceofaparticulardisadvantage(s)isdependentupon the nature of the investigation, the study population and themethod of data collection. Thesedisadvantagesincludethefollowing:

Astwosetsofdatamustbecollected,involvingtwocontactswiththestudypopulation,thestudyismoreexpensiveandmoredifficulttoimplement.Italsorequiresalongertimetocomplete,particularlyifyouareusinganexperimentaldesign,asyouwillneedtowaituntilyourinterventioniscompletedbeforeyoucollectthesecondsetofdata.Insomecasesthetimelapsebetweenthetwocontactsmayresultinattritioninthestudypopulation.Itispossiblethatsomeofthosewhoparticipatedinthepre-testmaymoveoutoftheareaorwithdrawfromtheexperimentforotherreasons.Oneofthemainlimitationsofthisdesign,initssimplestform,isthatasitmeasurestotalchange,youcannotascertainwhetherindependentorextraneousvariablesareresponsibleforproducingchangeinthedependentvariable.Also,itisnotpossibletoquantifythecontributionofindependentandextraneousvariablesseparately.Ifthestudypopulationisveryyoungandifthereisasignificanttimelapsebetweenthebefore-and-aftersetsofdatacollection,changesinthestudypopulationmaybebecauseitismaturing.Thisisparticularlytruewhenyouarestudyingyoungchildren.Theeffectofthismaturation,ifitissignificantlycorrelatedwiththedependentvariable,isreflectedatthe‘after’observationandisknownasthematurationeffect.Sometimestheinstrumentitselfeducatestherespondents.Thisisknownasthereactiveeffectoftheinstrument.Forexample,supposeyouwanttoascertaintheimpactofaprogrammedesignedtocreateawarenessofdrugsinapopulation.Todothis,youdesignaquestionnairelistingvariousdrugsandaskingrespondentstoindicatewhethertheyhaveheardofthem.Atthepre-teststagearespondent,whileansweringquestionsthatincludethenamesofthevariousdrugs,isbeingmadeawareofthem,andthiswillbereflectedinhis/herresponsesatthepost-teststage.Thus,theresearchinstrumentitselfhaseducatedthestudypopulationand,hence,hasaffectedthedependentvariable.Anotherexampleofthiseffectisastudydesignedtomeasuretheimpactofafamilyplanningeducationprogrammeonrespondents’awarenessofcontraceptivemethods.Moststudiesdesignedtomeasuretheimpactofaprogrammeonparticipants’awarenessfacethedifficultythatachangeinthelevelofawareness,tosomeextent,maybebecauseofthisreactiveeffect.Anotherdisadvantagethatmayoccurwhenyouusearesearchinstrumenttwicetogaugetheattitudeofapopulationtowardsanissueisapossibleshiftinattitudebetweenthetwopointsofdatacollection.Sometimespeoplewhoplacethemselvesattheextremepositionsofa

measurementscaleatthepre-teststagemay,foranumberofreasons,shifttowardsthemeanatthepost-teststage(seeFigure8.3).Theymightfeelthattheyhavebeentoonegativeortoopositiveatthepre-teststage.Therefore,themereexpressionofanattitudeinresponsetoaquestionnaireorinterviewhascausedthemtothinkaboutandaltertheirattitudeatthetimeofthepost-test.Thistypeofeffectisknownastheregressioneffect.

FIGURE8.3Theregressioneffect

Thelongitudinalstudydesign

Thebefore-and-afterstudydesignisappropriateformeasuringtheextentofchangeinaphenomenon,situation, problem, attitude, and so on, but is less helpful for studying the pattern of change. Todeterminethepatternofchangeinrelationtotime,alongitudinaldesignisused;forexample,whenyouwish tostudy theproportionofpeopleadoptingaprogrammeoveraperiod.Longitudinalstudiesarealso useful when you need to collect factual information on a continuing basis. You may want toascertainthetrendsinthedemandforlabour,immigration,changesintheincidenceofadiseaseorinthemortality,morbidityandfertilitypatternsofapopulation.Inlongitudinalstudiesthestudypopulationisvisitedanumberoftimesatregularintervals,usually

overalongperiod,tocollecttherequiredinformation(seeFigure8.4).Theseintervalsarenotfixedsotheirlengthmayvaryfromstudytostudy.Intervalsmightbeasshortasaweekorlongerthanayear.Irrespectiveofthesizeoftheinterval,thetypeofinformationgatheredeachtimeisidentical.Althoughthedatacollectedisfromthesamestudypopulation,itmayormaynotbefromthesamerespondents.Alongitudinalstudycanbeseenasaseriesofrepetitivecross-sectionalstudies.

FIGURE8.4ThelongitudinalstudydesignLongitudinal studies have many of the same disadvantages as before-and-after studies, in some

instancestoanevengreaterdegree.Inaddition,longitudinalstudiescansufferfromtheconditioningeffect.Thisdescribesasituationwhere,ifthesamerespondentsarecontactedfrequently,theybegintoknowwhatisexpectedofthemandmayrespondtoquestionswithoutthought,ortheymayloseinterest

intheenquiry,withthesameresult.Themainadvantageofalongitudinalstudyisthatitallowstheresearchertomeasurethepatternof

change and obtain factual information, requiring collection on a regular or continuing basis, thusenhancingitsaccuracy.

Studydesignsbasedonthereferenceperiod

Thereferenceperiod refers to the time-frame inwhichastudy isexploringaphenomenon,situation,eventorproblem.Studiesarecategorisedfromthisperspectiveas:

retrospective;prospective;retrospective–prospective.

Theretrospectivestudydesign

Retrospective studies investigate aphenomenon, situation,problemor issue thathashappened in thepast.Theyareusuallyconductedeitheronthebasisofthedataavailableforthatperiodoronthebasisof respondents’ recallof thesituation(Figure8.5a).Forexample, studiesconductedon the followingtopicsareclassifiedasretrospectivestudies:

ThelivingconditionsofAboriginalandTorresStraitIslanderpeoplesinAustraliaintheearlytwentiethcentury.TheutilisationoflandbeforetheSecondWorldWarinWesternAustralia.AhistoricalanalysisofmigratorymovementsinEasternEuropebetween1915and1945.Therelationshipbetweenlevelsofunemploymentandstreetcrime.

Theprospectivestudydesign

Prospective studies refer to the likely prevalence of a phenomenon, situation, problem, attitude oroutcomeinthefuture(Figure8.5b).Suchstudiesattempttoestablishtheoutcomeofaneventorwhatislikelytohappen.Experimentsareusuallyclassifiedasprospectivestudiesastheresearchermustwaitfor an intervention to register its effect on the study population. The following are classified asprospectivestudies:

Todetermine,underfieldconditions,theimpactofmaternalandchildhealthservicesonthelevelofinfantmortality.Toestablishtheeffectsofacounsellingserviceontheextentofmaritalproblems.Todeterminetheimpactofrandombreathtestingonthepreventionofroadaccidents.Tofindouttheeffectofparentalinvolvementonthelevelofacademicachievementoftheirchildren.TomeasuretheeffectsofachangeinmigrationpolicyontheextentofimmigrationinAustralia.

Theretrospective–prospectivestudydesign

Retrospective–prospectivestudiesfocusonpasttrendsinaphenomenonandstudyitintothefuture.Part of the data is collected retrospectively from the existing records before the intervention isintroducedandthenthestudypopulationisfollowedtoascertaintheimpactoftheintervention(Figure8.5c).

FIGURE8.5(a)Retrospectivestudydesign;(b)prospectivestudydesign;(c)retrospective–prospectivestudydesign.A study is classified under this categorywhen youmeasure the impact of an interventionwithout

havingacontrolgroup.Infact,mostbefore-and-afterstudies,ifcarriedoutwithouthavingacontrol–wherethebaselineisconstructedfromthesamepopulationbeforeintroducingtheintervention–willbeclassified as retrospective–prospective studies.Trend studies,which become the basis of projections,fallintothiscategorytoo.Someexamplesofretrospective–prospectivestudiesare:

Theeffectofrandombreathtestingonroadaccidents.Theimpactofincentivesontheproductivityoftheemployeesofanorganisation.Theimpactofmaternalandchildhealthservicesontheinfantmortalityrate.Theeffectofanadvertisementonthesaleofaproduct.

Studydesignsbasedonthenatureoftheinvestigation

Onthebasisofthenatureoftheinvestigation,studydesignsinquantitativeresearchcanbeclassifiedas:

experimental;non-experimental;quasi-orsemi-experimental.

To understand the differences, let us consider some examples. Suppose you want to test thefollowing: the impactof aparticular teachingmethodon the levelof comprehensionof students; theeffectiveness of a programme such as random breath testing on the level of road accidents; or theusefulnessofadrugsuchasazidothymidine(AZT)intreatingpeoplewhoareHIV-positive;orimagineanysimilarsituationinyourownacademicorprofessionalfield.Insuchsituationsthereisassumedtobeacause-and-effectrelationship.Therearetwowaysofstudyingthisrelationship.Thefirstinvolvesthe researcher (or someone else) introducing the intervention that is assumed to be the ‘cause’ ofchange,andwaitinguntilithasproduced–orhasbeengivensufficienttimetoproduce–thechange.Thesecondconsistsoftheresearcherobservingaphenomenonandattemptingtoestablishwhatcausedit. In this instance the researcher starts from the effect(s) or outcome(s) and attempts to determinecausation.Ifarelationshipisstudiedinthefirstway,startingfromthecausetoestablishtheeffects,itisclassifiedasanexperimentalstudy.Ifthesecondpathisfollowed–thatis,startingfromtheeffectstotracethecause–itisclassifiedasanon-experimentalstudy(seeFigure8.6).

FIGURE8.6Experimentalandnon-experimentalstudiesIntheformercasetheindependentvariablecanbe‘observed’,introduced,controlledormanipulated

bytheresearcherorsomeoneelse,whereas in the latter thiscannothappenas theassumedcausehasalreadyoccurred. Instead, theresearcher retrospectively links thecause(s) to theoutcome(s).Asemi-experimental study orquasi-experimental study has the properties of both experimental and non-experimentalstudies;partofthestudymaybenon-experimentalandtheotherpartexperimental.Anexperimentalstudycanbecarriedoutineithera‘controlled’ora‘natural’environment.Foran

experimentinacontrolledenvironment,theresearcher(orsomeoneelse)introducestheinterventionorstimulustostudyitseffects.Thestudypopulationisina‘controlled’situationsuchasaroom.Foranexperiment in a ‘natural’ environment, the study population is exposed to an intervention in its ownenvironment.Experimentalstudiescanbefurtherclassifiedonthebasisofwhetherornotthestudypopulationis

randomlyassigned todifferent treatmentgroups.Oneof thebiggestproblems in comparabledesigns(thoseinwhichyoucomparetwoormoregroups)isalackofcertaintythatthedifferentgroupsarein

fact comparable in every respect except the treatment. The process of randomisation is designed toensure that the groups are comparable. In a randomdesign, the study population, the experimentaltreatmentsorbotharenotpredeterminedbutrandomlyassigned(seeFigure8.7).Randomassignmentin experiments means that any individual or unit of a study population group has an equal andindependent chanceofbecomingpartof anexperimentalor controlgroupor, in thecaseofmultipletreatmentmodalities,anytreatmenthasanequalandindependentchanceofbeingassignedtoanyofthepopulationgroups.Itisimportanttonotethattheconceptofrandomisationcanbeappliedtoanyoftheexperimentaldesignswediscuss.

FIGURE8.7Randomisationinexperiments

Experimentalstudydesigns

Therearesomanytypesofexperimentaldesignthatnotallofthemcanbeconsideredwithinthescopeofthisbook.Thissection,therefore,isconfinedtodescribingthosemostcommonlyusedinthesocialsciences, the humanities, public health,marketing, education, epidemiology, socialwork, and so on.Thesedesignshavebeencategorisedas:

theafter-onlyexperimentaldesign;thebefore-and-afterexperimentaldesign;thecontrolgroupdesign;thedouble-controldesign;thecomparativedesign;the‘matchedcontrol’experimentaldesign;theplacebodesign.

FIGURE8.8Theafter-onlydesign

Theafter-onlyexperimentaldesignIn an after-only design the researcher knows that a population is being, or has been, exposed to aninterventionandwishes tostudy its impacton thepopulation. In thisdesign, informationonbaseline(pre-test or before observation) is usually ‘constructed’ on the basis of respondents’ recall of thesituationbeforetheintervention,orfrominformationavailableinexistingrecords–secondarysources(Figure8.8).Thechangeinthedependentvariableismeasuredbythedifferencebetweenthe‘before’(baseline)and‘after’datasets.Technically,thisisaveryfaultydesignformeasuringtheimpactofaninterventionastherearenoproperbaselinedatatocomparethe‘after’observationwith.Therefore,oneof the major problems of this design is that the two sets of data are not strictly comparable. Forexample,someofthechangesinthedependentvariablemaybeattributabletothedifferenceinthewaythetwosetsofdatawerecompiled.Anotherproblemwiththisdesignisthatitmeasurestotalchange,including change attributable to extraneous variables; hence, it cannot identify the net effect of anintervention.However, this design iswidely used in impact assessment studies, as in real lifemanyprogrammes operate without the benefit of a planned evaluation at the programme planning stage(though this is fast changing) in which case it is just not possible to follow the sequence strictly –collectionofbaselineinformation,implementationoftheprogrammeandthenprogrammeevaluation.Anevaluatorthereforehasnochoicebuttoadoptthisdesign.Inpractice,theadequacyofthisdesigndependsonhavingreasonablyaccuratedataavailableabout

the prevalence of a phenomenon before the intervention is introduced. This might be the case forsituations such as the impact of random breath testing on road accidents, the impact of a healthprogrammeonthemortalityofapopulation,theimpactofanadvertisementonthesaleofaproduct,theimpactofadeclineinmortalityonthefertilityofapopulation,ortheimpactofachangeinimmigrationpolicy on the extent of immigration. In these situations it is expected that accurate records are keptaboutthephenomenonunderstudyandsoitmaybeeasiertodeterminewhetheranychangeintrendsisprimarilybecauseoftheintroductionoftheinterventionorchangeinthepolicy.

Thebefore-and-afterexperimentaldesignThe before-and-after design overcomes the problem of retrospectively constructing the ‘before’observationbyestablishingitbeforetheinterventionisintroducedtothestudypopulation(seeFigure8.2).Then,whentheprogrammehasbeencompletelyimplementedorisassumedtohavehaditseffecton the population, the ‘after’ observation is carried out to ascertain the impact attributable to the

intervention(seeFigure8.9).

FIGURE8.9Measurementofchangethroughabefore-and-afterdesignThe before-and-after design takes care of only one problem of the after-only design – that is, the

comparability of the before-and-after observations. It still does not enable one to conclude that anychange– inwholeor inpart– canbeattributed to theprogramme intervention.Toovercome this, a‘control’groupisused.Before-and-afterdesignsmayalsosufferfromtheproblemsidentifiedearlierinthis chapter in the discussion of before-and-after study designs. The impact of the intervention inbefore-and-afterdesigniscalculatedasfollows:

[changeindependentvariable]=

[statusofthedependentvariableatthe‘after’observation]–[statusofthedependentvariableatthe‘before’observation]

ThecontrolgroupdesignInastudyutilisingthecontrolgroupdesigntheresearcherselectstwopopulationgroupsinsteadofone:a control group and an experimental group (Figure 8.10). These groups are expected to becomparable as far as possible in every respect except for the intervention (that is assumed to be thecauseresponsibleforbringingaboutthechange).Theexperimentalgroupeitherreceivesorisexposedtotheintervention,whereasthecontrolgroupisnot.Firstly,the‘before’observationsaremadeonbothgroups at the same time. The experimental group is then exposed to the intervention. When it isassumed that the interventionhashadan impact,an ‘after’observation ismadeonbothgroups.Anydifference in the ‘before’ and ‘after’ observations between the groups regarding the dependentvariable(s)isattributedtotheintervention.

FIGURE8.10Thecontrolexperimentaldesign

Intheexperimentalgroup,totalchangeinthedependentvariable(Ye)canbecalculatedasfollows:

Ye=(Y″e–Y'e)

where

Y″e=‘after’observationontheexperimentalgroupY'e=‘before’observationontheexperimentalgroup

Inotherwords,

(Y″e–Y'e)= (impactofprogrammeintervention)±(impactofextraneousvariables)±(impactofchancevariables)

Inthecontrolgroup,totalchangeinthedependentvariable(Yc)canbecalculatedasfollows:

Yc=(Y″c–Y'c)

where

Y″c=post-testobservationonthecontrolgroupY'c=pre-testobservationonthecontrolgroup

Inotherwords,

(Y″c–Y'c)= (impactofextraneousvariables)±(impactofchancevariables)

Thedifferencebetweenthecontrolandexperimentalgroupscanbecalculatedas

(Y″e–Y'e)–(Y″c–Y'c),

whichis

{(impactofprogrammeintervention)±(impactofextraneousvariablesinexperimentalgroups)±(impactofchancevariablesinexperimentalgroups)}-{(impactofextraneousvariablesincontrolgroup)±(impactofchancevariablesincontrolgroup)}

Usingsimplearithmeticoperations,thisequalstheimpactoftheintervention.

Therefore, the impact of any intervention is equal to the difference in the ‘before’ and ‘after’observationsinthedependentvariablebetweentheexperimentalandcontrolgroups.

Itisimportanttorememberthatthechiefobjectiveofthecontrolgroupistoquantifytheimpactofextraneousvariables.Thishelpsyoutoascertaintheimpactoftheinterventiononly.

Thedouble-controldesignAlthough the control design helps you to quantify the impact that can be attributed to extraneousvariables,itdoesnotseparateoutothereffectsthatmaybeduetotheresearchinstrument(suchasthereactiveeffect)or respondents(suchas thematurationorregressioneffects,orplaceboeffect).Whenyouneedtoidentifyandseparateouttheseeffects,adouble-controldesignisrequired.Indouble-controlstudies,youhavetwocontrolgroupsinsteadofone.Toquantify,say,thereactive

effect of an instrument, you excludeoneof the control groups from the ‘before’ observation (Figure8.11).

FIGURE8.11Double-controldesigns

Youcancalculatethedifferenteffectsasfollows:

(Y″e–Y'e)=(impactofprogrammeintervention)±(impactofextraneousvariables)±(reactiveeffect)±(randomeffect)(Y″c1–Y'c1)=(impactofextraneousvariables)±(reactiveeffect)±(randomeffect)(Y″c2–Y'c1)=(impactofextraneousvariables)±(randomeffect)

(Notethat(Y″c2–Y'c1)andnot(Y″c2–Y'c2)asthereisno‘before’observationforthesecondcontrolgroup.)

(Y'e–Y'e)–(Y″c1–Y'c1)=impactofprogrammeintervention(Y″c1–Y'c1)–(Y'c2–Y'c1)=reactiveeffect

Theneteffectof theprogrammeinterventioncanbecalculatedinthesamemannerasforthecontrolgroupdesignsasexplainedearlier.

Thecomparativedesign

Sometimes you seek to compare the effectiveness of different treatment modalities and in suchsituationsacomparativedesignisappropriate.With a comparative design, as with most other designs, a study can be carried out either as an

experiment or as a non-experiment. In the comparative experimental design, the study population isdividedintothesamenumberofgroupsas thenumberof treatments tobetested.Foreachgroupthebaselinewithrespecttothedependentvariableisestablished.Thedifferenttreatmentmodelsarethenintroducedtothedifferentgroups.Afteracertainperiod,whenitisassumedthatthetreatmentmodelshave had their effect, the ‘after’ observation is carried out to ascertain any change in the dependentvariable. The degree of change in the dependent variable in the different population groups is thencomparedtoestablishtherelativeeffectivenessofthevariousinterventions.Inthenon-experimentalformofcomparativedesign,groupsalreadyreceivingdifferentinterventions

are identified,andonly thepost-observationwithrespect to thedependentvariable isconducted.Thepre-testdatasetisconstructedeitherbyaskingthestudypopulationineachgrouptorecalltherequiredinformation relating to the period before the introduction of the treatment, or by extracting suchinformation from existing records. Sometimes a pre-test observation is not constructed at all, on theassumptionthatifthegroupsarecomparablethebaselinemustbeidentical.Aseachgroupisassumedtohave thesamebaseline, thedifference in thepost-testobservation isassumed tobebecauseof theintervention.Toillustratethis,imagineyouwanttocomparetheeffectivenessofthreeteachingmodels(A,Band

C)onthelevelofcomprehensionofstudentsinaclass(Figure8.12).Toundertakethestudy,youdividetheclassintothreegroups(X,YandZ),throughrandomisation,toensuretheircomparability.Beforeexposingthesegroupstotheteachingmodels,youfirstestablishthebaselineforeachgroup’slevelofcomprehensionofthechosensubject.Youthenexposeeachgrouptoadifferentteachingmodeltoteachthechosensubject.Afterwards,youagainmeasurethegroups’levelsofcomprehensionofthematerial.SupposeXaistheaveragelevelofcomprehensionofgroupXbeforethematerialistaught,andXa'isthis group’s average level of comprehension after thematerial is taught. The change in the level ofcomprehension,Xa'–XaisthereforeattributedtomodelA.Similarly,changesingroupYandZ,Yb'–YbandZc'–Zc, are attributed to teachingmodelsB andC respectively.The changes in the averagelevelofcomprehensionforthethreegroupsarethencomparedtoestablishwhichteachingmodelisthemost effective. (Note that extraneous variables will affect the level of comprehension in all groupsequally,astheyhavebeenformedrandomly.)

FIGURE8.12ComparativeexperimentaldesignItisalsopossibletosetupthisstudyasanon-experimentalone,simplybyexposingeachgroupto

oneofthethreeteachingmodels,followingupwithan‘after’observation.Thedifferenceinthelevelsofcomprehensionisattributedto thedifferencein theteachingmodelsas it isassumedthat thethreegroupsarecomparablewithrespecttotheiroriginallevelofcomprehensionofthetopic.

ThematchedcontrolexperimentaldesignComparative groups are usually formed on the basis of their overall comparabilitywith respect to arelevantcharacteristicinthestudypopulation,suchassocioeconomicstatus,theprevalenceofacertaincondition or the extent of a problem in the study population. In matched studies, comparability isdeterminedon an individual-by-individual basis.Two individuals from the study populationwho arealmostidenticalwithrespecttoaselectedcharacteristicand/orcondition,suchasage,genderortypeofillness,arematchedandtheneachisallocatedtoaseparategroup(thematchingisusuallydoneonaneasilyidentifiablecharacteristic).Inthecaseofamatchedcontrolexperiment,oncethetwogroupsareformed,youasaresearcherdecidethroughrandomisationorotherwisewhichgroupistobeconsideredcontrol,andwhichexperimental.Thematcheddesigncanposeanumberofchallenges:

Matchingincreasesindifficultywhencarriedoutonmorethanonevariable.Matchingonvariablesthatarehardtomeasure,suchasattitudeoropinion,isextremelydifficult.Sometimesitishardtoknowwhichvariabletochooseasabasisformatching.Youmaybeabletobaseyourdecisionuponpreviousfindingsoryoumayhavetoundertakeapreliminarystudytodetermineyourchoiceofvariable.

Matchedgroupsaremostcommonlyusedinthetestingofnewdrugs.

The‘placebo’designApatient’sbeliefthats/heisreceivingtreatmentcanplayanimportantroleinhis/herrecoveryfromanillness even if treatment is ineffective. This psychological effect is known as theplacebo effect. Aplacebodesignattempts todeterminetheextentof thiseffect.Aplacebostudy involves twoor threegroups,dependingonwhetherornottheresearcherwantstohaveacontrolgroup(Figure8.13).Iftheresearcherdecides tohaveacontrolgroup, thefirstgroupreceives the treatment, thesecondreceivestheplacebo treatment and the third– thecontrolgroup– receivesnothing.Thedecisionas towhichgroupwillbethetreatment,theplaceboorthecontrolgroupcanalsobemadethroughrandomisation.

FIGURE8.13Theplacebodesign

Otherdesignscommonlyusedinquantitativeresearch

Therearesomeresearchdesignsthatmaybeclassifiedinthetypologydescribedabovebut,becauseoftheir uniqueness and prevalence, have acquired their own names. They are therefore describedseparatelybelow.

Thecross-overcomparativeexperimentaldesign

Thedenialoftreatmenttothecontrolgroupisconsideredunethicalbysomeprofessionals.Inaddition,thedenialoftreatmentmaybeunacceptabletosomeindividualsinthecontrolgroup,whichcouldresultin them dropping out of the experiment and/or going elsewhere to receive treatment. The formerincreases ‘experimental mortality’ and the latter may contaminate the study. The cross-overcomparative experimental design makes it possible to measure the impact of a treatment withoutdenyingtreatmenttoanygroup,thoughthisdesignhasitsownproblems.Inthecross-overdesign,alsocalledtheABABdesign(Grinnell1993:104),twogroupsareformed,

theinterventionisintroducedtooneofthemand,afteracertainperiod,theimpactofthisinterventionismeasured.Thentheinterventionsare‘crossedover’;thatis,theexperimentalgroupbecomesthecontroland vice versa, sometimes repeatedly over the period of the study (Figure 8.14). However, in thisdesign, population groups do not constitute experimental or control groups but only segments uponwhichexperimentalandcontrolobservationsareconducted.

FIGURE8.14Thecross-overexperimentaldesignOneofthemaindisadvantagesofthisdesignisdiscontinuityintreatment.Themainquestionis:what

impactwouldinterventionhaveproducedhaditnotbeenprovidedinsegments?

Thereplicatedcross-sectionaldesign

In practice one usually examines programmes already in existence and ones in which clients are atdifferent stages of an intervention. Evaluating the effectiveness of such programmes within aconventional experimental design is impossible because a baseline cannot be established as theintervention has already been introduced. In this situation, the usualmethod of selecting a group of

peoplewhowererecentlyrecruitedtotheprogrammeandfollowingthemthroughuntiltheinterventionhasbeencompletedmaytakealongtime.Insuchsituations,itispossibletochooseclientswhoareatdifferentphasesoftheprogrammetoformthebasisofyourstudy(Figure8.15).

FIGURE8.15Thereplicatedcross-sectionaldesignThis design is based upon the assumption that participants at different stages of a programme are

similarintermsoftheirsocioeconomic–demographiccharacteristicsandtheproblemforwhichtheyareseekingintervention.Assessmentoftheeffectivenessofaninterventionisdonebytakingasampleofclientsatdifferentstagesoftheintervention.Thedifferenceinthedependentvariableamongclientsatintakeandterminationstageisconsideredtobetheimpactoftheintervention.

Trendstudies

Ifyouwanttomapchangeoveraperiod,atrendstudyisthemostappropriatemethodofinvestigation.Trendanalysisenablesyoutofindoutwhathashappenedinthepast,whatishappeningnowandwhatislikelytohappeninthefutureinapopulationgroup.Thisdesigninvolvesselectinganumberofdataobservationpointsinthepast,togetherwithapictureofthepresentorimmediatepastwithrespecttothephenomenonunderstudy,andthenmakingcertainassumptionsastofuturetrends.Inawayyouarecollectingcross-sectionalobservationsabout the trendbeingobservedatdifferentpoints in timeoverpast–present–future.Fromthesecross-sectionalobservationsyoudrawconclusionsaboutthepatternofchange.Trendstudiesareusefulinmakingforecastingbyextrapolatingpresentandpasttrendsthusmakinga

valuablecontributiontoplanning.Trendsregardingthephenomenonunderstudycanbecorrelatedwithother characteristics of the study population. For example, youmaywant to examine the changes inpoliticalpreferenceofastudypopulationinrelationtoage,gender,incomeorethnicity.Thisdesigncanalsobeclassifiedasretrospective–prospectivestudyonthebasisofthereferenceperiodclassificationsystemdevelopedearlierinthischapter.

Cohortstudies

Cohort studies are based upon the existence of a common characteristic such as year of birth,graduationormarriage,withinasubgroupofapopulation.Supposeyouwanttostudytheemploymentpattern of a batch of accountants who graduated from a university in 1975, or study the fertilitybehaviour of women who weremarried in 1930. To study the accountants’ career paths you wouldcontact all the accountantswho graduated from the university in 1975 to find out their employmenthistories.Similarly,youwould investigate the fertilityhistoryof thosewomenwhomarried in1930.Bothofthesestudiescouldbecarriedouteitherascross-sectionalorlongitudinaldesigns.Ifyouadopta cross-sectional design you gather the required information in one go, but if you choose thelongitudinaldesignyoucollecttherequiredinformationatdifferentpointsintimeoverthestudyperiod.Boththesedesignshavetheirstrengthsandweaknesses. In thecaseofa longitudinaldesign, it isnotimportant for the required information to be collected from the same respondents; however, it isimportantthatalltherespondentsbelongtothecohortbeingstudied;thatis,intheaboveexamplestheymusthavegraduatedin1975ormarriedin1930.

Panelstudies

Panelstudiesaresimilartotrendandcohortstudiesexceptthatinadditiontobeinglongitudinaltheyarealsoprospective innatureand the information isalwayscollectedfromthesamerespondents. (Intrend and cohort studies the information can be collected in a cross-sectional manner and theobservationpointscanbe retrospectivelyconstructed.)Supposeyouwant tostudy thechanges in thepatternofexpenditureonhouseholditemsinacommunity.Todothis,youwouldselectafewfamiliestofindouttheamounttheyspendeveryfortnightonhouseholditems.Youwouldkeepcollectingthesame information from the same families over a period of time to ascertain the changes in theexpenditurepattern.Similarly,apanelstudydesigncouldbeused tostudy themorbiditypattern inacommunity.

Blindstudies

Theconceptofablindstudy canbeusedwithcomparableandplaceboexperimentaldesignsand isappliedtostudiesmeasuringtheeffectivenessofadrug.Inablindstudy,thestudypopulationdoesnotknowwhether it isgetting realor fake treatmentorwhich treatmentmodality.Themainobjectiveofdesigningablindstudyistoisolatetheplaceboeffect.

Double-blindstudies

Theconceptofadouble-blindstudyisverysimilartothatofablindstudyexceptthatitalsotriestoeliminate researcherbiasbyconcealing the identityof theexperimentalandplacebogroups from theresearcher. In other words, in a double-blind study neither the researcher nor the study participantsknowwho is receiving real andwho is receiving fake treatment orwhich treatmentmodel they arereceiving.

Studydesignsinqualitativeresearch

Thissectionprovidesabriefdescriptionofsomeofthecommonlyuseddesignsinqualitativeresearch.

Foranin-depthunderstandingyouareadvisedtoconsultbooksonqualitativeresearch.

Casestudy

Thecasestudy,thoughdominantlyaqualitativestudydesign,isalsoprevalentinquantitativeresearch.Acasecouldbeanindividual,agroup,acommunity,aninstance,anepisode,anevent,asubgroupofapopulation,atownoracity.Tobecalledacasestudyitisimportanttotreatthetotalstudypopulationasoneentity.Inacasestudydesign the ‘case’youselectbecomes thebasisofa thorough,holisticand in-depth

exploration of the aspect(s) that youwant to find out about. It is an approach ‘inwhich a particularinstance or a few carefully selected cases are studied intensively’ (Gilbert 2008: 36). According toBurns(1997:364),‘toqualifyasacasestudy,itmustbeaboundedsystem,anentityinitself.Acasestudyshouldfocusonaboundedsubject/unitthatiseitherveryrepresentativeorextremelyatypical.’Acase study according to Grinnell (1981: 302), ‘is characterized by a very flexible and open-endedtechniqueofdatacollectionandanalysis’.Thecasestudydesignisbasedupontheassumptionthatthecasebeingstudiedisatypicalofcasesof

acertaintypeandthereforeasinglecasecanprovideinsightintotheeventsandsituationsprevalentinagroupfromwherethecasehasbeendrawn.AccordingtoBurns(1997:365),‘Inacasestudythefocusof attention is the case in its idiosyncratic complexity, not on the whole population of cases.’ Inselecting a case therefore you usually use purposive, judgemental or information-oriented samplingtechniques.Itisaveryusefuldesignwhenexploringanareawherelittleisknownorwhereyouwanttohavea

holisticunderstandingofthesituation,phenomenon,episode,site,grouporcommunity.Thisdesignisofimmenserelevancewhenthefocusofastudyisonextensivelyexploringandunderstandingratherthan confirming and quantifying. It provides an overview and in-depth understanding of a case(s),processandinteractionaldynamicswithinaunitofstudybutcannotclaimtomakeanygeneralisationstoapopulationbeyondcasessimilartotheonestudied.Inthisdesignyourattemptisnottoselectarandomsamplebutacasethatcanprovideyouwithas

much information as possible to understand the case in its totality.When studying an episode or aninstance, you attempt to gather information from all available sources so as to understand it in itsentirety.Ifthefocusofyourstudyisagrouporcommunityyoushouldspendsufficienttimebuildingatrustworthyrapportwithitsmembersbeforecollectinganyinformationaboutthem.Though you can use a singlemethod, the use ofmultiplemethods to collect data is an important

aspect of a case study, namely in-depth interviewing, obtaining information from secondary records,gatheringdatathroughobservations,collectinginformationthroughfocusgroupsandgroupinterviews,etc.However, it is importantthatat thetimeofanalysisyoucontinuetoconsiderthecaseasasingleentity.

Oralhistory

Oralhistoryismoreamethodofdatacollectionthanastudydesign;however,inqualitativeresearch,thishasbecomeanapproach tostudyperceptions,experiencesandaccountsofaneventorgatheringhistoricalknowledgeasviewedby individuals. It isapictureof something insomeone’sownwords.Oral history is a process of obtaining, recording, presenting and interpreting historical or currentinformation,baseduponpersonalexperiencesandopinionsofsomemembersofastudygrouporunit.Theseopinionsorexperiencescouldbebaseduponeye-witnessevidenceorinformationpassedonfrom

other sources such as older people, ancestors, folklore, stories. According to Ritchie (2003: 19),‘Memoryisthecoreoforalhistory,fromwhichmeaningcanbeextractedandpreserved.Simplyput,oralhistorycollectsmemories andpersonal commentariesofhistorical significance through recordedinterviews.’AccordingtoBurns(1997:368),‘theseareusuallyfirstpersonnarrativesthattheresearchercollectsusingextensiveinterviewingofasingleindividual’.Intermsofdesignitisquitesimple.Youfirstdecidewhattypesofaccount,experience,perceptionor

historicaleventyouwanttofindoutabout.Thenyouneedtoidentifytheindividualsorsources(whichcouldbedifficultandtimeconsuming)thatcanbestprovideyouwiththeneededinformation.Youthencollectinformationfromthemtobeanalysedandinterpreted.

Focusgroups/groupinterviews

Focusgroupsareaformofstrategyinqualitativeresearchinwhichattitudes,opinionsorperceptionstowards an issue, product, service or programme are explored through a free and open discussionbetweenmembersofagroupandtheresearcher.Bothfocusgroupsandgroupinterviewsarefacilitatedgroupdiscussionsinwhicharesearcherraisesissuesorasksquestionsthatstimulatediscussionamongmembersofthegroup.Becauseofitslowcost,itisapopularmethodforfindinginformationinalmosteveryprofessionalareaandacademicfield.Social,politicalandbehaviouralscientists,marketresearchandproducttestingagencies,andurbanandtownplanningexpertsoftenusethisdesignforavarietyofsituations.Forexample,inmarketingresearchthisdesigniswidelyusedtofindoutconsumers’opinionofandfeedbackonaproduct, theiropinionson thequalityof theproduct, itsacceptanceandappeal,priceandpackaging,howtoimprovethequalityandincreasethesaleoftheproduct,etc.Focusgroupsarealsoprevalentinformativeandsummativeevaluationsandfordevelopingsocialprogrammesandservices.Itisalsoausefultoolinsocialandurbanplanningforidentifyingissues,options,developmentstrategies,andfutureplanninganddevelopmentdirections.Initsdesignit isverysimple.Youasaresearcherselectagroupofpeoplewhoyouthinkarebest

equipped to discusswhat youwant to explore. The group could comprise individuals drawn from agroup of highly trained professionals or average residents of a community depending upon theobjectivesof thefocusgroup.In theformationofafocusgroupthesizeof thegroupisanimportantconsideration.Itshouldbeneithertoolargenortoosmallasthiscanimpedeupontheextentandqualityofthediscussion.Approximatelyeighttotenpeoplearetheoptimalnumberforsuchdiscussiongroups.Youalsoneedtoidentifycarefullytheissuesfordiscussionprovidingeveryopportunityforadditionalrelevantones toemerge.Asa researcheryoualsoneed todecide, inconsultationwith thegroup, theprocess of recording the discussion. This may include fixing the times that the group can meet toextensivelydiscussingthe issuesandarrivingatagreementsonthem.Yourrecordsof thediscussionsthenbecome thebasisofanalysis for findingsandconclusions.Themaindifferencebetweena focusgroupandagroupinterviewisinthedegreeofspecificitywithrespecttotheissuestobediscussed.Theissuesdiscussed in focusgroupsaremorespecificandfocused than ingroup interviewsand theyarelargely predetermined by the researcher. In a group interview you let the group members discusswhatevertheywant.However,yourroleasaresearcheristobringthembacktotheissuesofinterestasidentifiedbythegroup.Compared with other designs this is less expensive and needs far less time to complete. The

information generated can be detailed and rich and can be used to explore a vast variety of issues.However,thedisadvantageisthatifthediscussionisnotcarefullydirecteditmayreflecttheopinionofthosewhohaveatendencytodominateagroup.Thisdesignisveryusefulforexploringthediversityinopinionsondifferentissuesbutwillnothelpyouifyouwanttofindouttheextentormagnitudeofthis

diversity.

Participantobservation

Participantobservation is another strategy for gathering information about a social interaction or aphenomenoninqualitativestudies.Thisisusuallydonebydevelopingacloseinteractionwithmembersof a group or ‘living’ in the situation which is being studied. Though predominantly a qualitativeresearchdesign,itisalsousedinquantitativeresearch,dependinguponhowtheinformationhasbeengenerated and recorded. In qualitative research, an observation is always recorded in a descriptiveformatwhereasinquantitativeresearchitisrecordedeitherincategoriesoronascale.Itcanalsobeacombinationofboth– somecategorisationand somedescriptionor categorisationaccompaniedbyadescriptive explanation. You can also change a descriptive recording into a categorical one throughanalysis and classification. In addition to the observation itself, where you as an observer generateinformation,theinformationcanalsobecollectedthroughothermethodssuchasinformalinterviewing,in-depthinterviewing,groupdiscussions,previousdocuments,oralhistories.Useofmultiplemethodswillenhancetherichnessoftheinformationcollectedbyparticipantobservation.In its design it is simple.You as a researcher get involved in the activities of the group, create a

rapport with group members and then, having sought their consent, keenly observe the situation,interaction,siteorphenomenon.Youmakedetailednotesofwhatyouobserveinaformatthatbestsuitsyouaswellasthesituation.Youcanalsocollectinformationusingothermethodsofdatacollection,ifneedbe.Youanalyserecordsofyourobservationsanddatacollectedbyothermeanstodrawinferencesandconclusions.Themainadvantageofparticipantobservationisthatasyouspendsufficienttimewiththegroupor

inthesituation,yougainmuchdeeper,richerandmoreaccurateinformation,butthemaindisadvantageisthat,ifyouarenotverycareful,youcanintroduceyourownbias.

Holisticresearch

Theholisticapproachtoresearchisonceagainmoreaphilosophythanastudydesign.Thedesignisbaseduponthephilosophythatasamultiplicityoffactorsinteractsinourlives,wecannotunderstandaphenomenonfromjustoneortwoperspectives.Tounderstandasituationorphenomenonyouneedtolookatitinitstotality–thatis,holisticallyfromeveryperspective.Youcanuseanydesignwhenexploringasituationfromdifferentperspectivesandtheuseofmultiple

methodsisprevalentanddesirable.

Communitydiscussionforums

Communitydiscussionforumsaredesignedtofindopinions,attitudesand/orideasofacommunitywithregardtocommunityissuesandproblems.Itisoneoftheverypopularwaysofseekingacommunity’sparticipationindecidingaboutissuesofconcerntomembersofthecommunity.Suchforumsarealsoused for a variety of other reasons such as developing townplanning options and community healthprogrammesforacommunity,seekingparticipationofitsmembersinresolvingissuesrelatingtotrafficmanagement, infrastructure development and determining future directions for the area, informingcommunitiesofnewinitiatives.Communityforumsareverysimilar togroupdiscussionsexcept that theseareonabiggerscale in

terms of number of participants. Also, in group discussions youmay select the participants, but forcommunity forums there is self-selection of the participants as they are open to everyone with aninterestintheissuesorconcerns.Theresearcherusuallyuseslocalmediatoinformtheresidentsofalocalcommunityabouttheforums.This is a useful design to find out the spread of issues, concerns, etc., at a community level. It is

economicalandquickbuttherearesomedisadvantages.Forexample, it ispossiblethatafewpeoplewithavestedinterestcandominatethediscussioninaforumanditisequallypossiblethatonoccasionsthere may be very low attendance. Such situations may result in the discussion not reflecting thecommunityattitudes.

Reflectivejournallog

Basically,thisdesignentailskeepingareflectivejournallogofyourthoughtsasaresearcherwheneveryou notice anything, talk to someone, participate in an activity or observe something that helps youunderstandoraddtowhateveryouaretryingtofindoutabout.Thesereflectiverecordsthenbecomethebasisofyour findingsandconclusions.Youcanhavea reflective journalas theonlymethodofdatacollectionoritcanbeusedincombinationwithothermethodssuchasinterviewing,groupinterviews,orsecondarysources.

Othercommonlyusedphilosophy-guideddesigns

Thereareanumberofotherapproachestoresearchthathaveacquiredrecognition,intermsofdesignandname,intheresearchliterature.Whilenotdesignsperse,theydoenhanceaparticularphilosophicalperspectiveinsocialresearch.Theseare:actionresearch,feministresearch,participatoryresearchandcollaborative enquiry. Strictly speaking, a piece of research within each of these could be eitherquantitativeorqualitative,thoughbymanytheyareconsidereddominantlyasqualitativedesigns.Theneed to place them in a separate category stems from their prominence and possible use in eachparadigm.Thesedesignsaremorephilosophyguidedthanmethodsbased.Forexample,actionresearchisguidedby thephilosophy thatapieceof researchshouldbefollowedbysomeformofappropriateactiontoachievebettermentinlifeorservice,andfeministresearchisinfluencedbythephilosophythatopposesandchallengesthedominantmalebiasinsocialscienceresearch;itseemstobelievethatissuesrelatingtowomenarebestunderstoodandresearchedbywomenalone.Forparticipatoryresearchandcollaborativeenquiry,theinvolvementofresearchparticipantsorthecommunityintheresearchprocessistheunderlyingphilosophy.Oneoftheimportantaspectsofallthese‘designs’isthattheyattempttoinvolveresearchparticipantsintheresearchprocess.Theresearchfindingsarethenusedtodepictthecurrentsituationwithrespecttocertainissuesorproblemsandhelptoformasoundbasisforstrategydevelopmenttodealwiththem.

Actionresearch

As the name suggests,action research comprises two components: action and research (see Figure8.16).Research is ameans toaction, either to improveyourpracticeor to takeaction todealwithaproblemoranissue.Sinceactionresearchisguidedbythedesiretotakeaction,strictlyspeakingitisnotadesignperse.Mostactionresearchisconcernedwithimprovingthequalityofservice.Itiscarriedouttoidentifyareasofconcern,developandtestalternatives,andexperimentwithnewapproaches.

FIGURE8.16ActionresearchdesignActionresearchseemstofollowtwotraditions.TheBritishtraditiontendstoviewactionresearchas

a means of improvement and advancement of practice (Carr & Kemmis 1986), whereas in the UStraditionitisaimedatsystematiccollectionofdatathatprovidesthebasisforsocialchange(Bogdan&Biklen1992).Actionresearch,incommonwithparticipatoryresearchandcollaborativeenquiry,isbasedupon

a philosophy of community development that seeks the involvement of community members.Involvement and participation of a community, in the total process from problem identification toimplementation of solutions, are the two salient features of all three approaches (action research,participatory research and collaborative enquiry). In all three, data is collected through a researchprocess, and changes are achieved through action. This action is taken either by officials of aninstitutionorthecommunityitselfinthecaseofactionresearch,orbymembersofacommunityinthecaseofcollaborativeorparticipatoryresearch.Therearetwofocusesofactionresearch:

1. Anexistingprogrammeorinterventionisstudiedinordertoidentifypossibleareasofimprovementintermsofenhancedefficacyand/orefficiency.Thefindingsbecomethebasisofbringingaboutchanges.

2. Aprofessionalidentifiesanunattendedproblemorunexplainedissueinthecommunityoramongaclientgroupandresearchevidenceisgatheredtojustifytheintroductionofanewserviceorintervention.Researchtechniquesestablishtheprevalenceoftheproblemortheimportanceofanissuesothatappropriateactioncanbetakentodealwithit.

Feministresearch

Feministresearchischaracterisedbyitsfeministtheoryphilosophicalbasethatunderpinsallenquiriesandfeministconcernsactastheguidingframework.Feministresearchdiffersfromtraditionalresearchinthreeways:

1. Itsmainfocusistheexperiencesandviewpointsofwomen.Itusesresearchmethodsaimedatexploringthese.

2. Itactivelytriestoremoveorreducethepowerimbalancebetweentheresearcherandrespondents.

3. Thegoaloffeministresearchischangingthesocialinequalitybetweenmenandwomen.Infact,feministresearchmaybeclassifiedasactionresearchintheareaofgenderinequality,usingresearchtechniquestocreateawarenessofwomen’sissuesandconcerns,andtofosteractionpromotingequalitybetweensexes.

Anystudydesigncouldbeusedinfeministresearch.

Participatoryandcollaborativeresearchenquiry

As alreadymentioned, to the author’smind, these are not designs per se but signify a philosophicalperspective that advocates the active involvement of research participants in the research process.Participatoryresearchisbasedupontheprincipleofminimisingthe‘gap’betweentheresearcherandthe research participants and increased community involvement and participation to enhance therelevanceoftheresearchfindingstotheirneeds.Itisassumedthatsuchinvolvementwillincreasethepossibility of the community accepting the research findings and, if need be, its willingness andinvolvement in solving the problems and issues that confront it.You canundertake a quantitative orqualitativestudyintheseenquiriesbutthemainemphasisisonpeople’sengagement,collaborationandparticipationintheresearchprocess.Inawaythesedesignsarebasedonthecommunitydevelopmentmodelwhere engagement of a community byway of consultation and participation in planning andexecutionof research tasks is imperative. In thesedesignsyouarenotmerelya researcherbutalsoacommunityorganiserseekingactiveparticipationofthecommunity.As a researcher you work at two different aspects: (1) community organisation and (2) research.

Through community organisationyou seek a community’s involvement andparticipation in planningandexecutionoftheresearchtasksandshareresearchfindingswithitsmembers.Intermsofresearch,your main responsibility is to develop, in consultation with the community, the research tasks andprocedures.Consultationwithresearchparticipantsisacontinuousandintegralpartofthesedesigns.

SummaryInthischaptervariousstudydesignsinbothquantitativeandqualitativeresearchhavebeenexamined.Foreachstudydesign,detailshavebeenprovidedon the situations inwhich thedesign is appropriate touse, its strengthsandweaknesses, and theprocessyouadoptinitsoperationalisation.Inquantitativeresearchthevariousstudydesignshavebeenexaminedfromthreeperspectives.Theterminologyusedtodescribe

theseperspectives is thatof theauthorbut thenamesof thestudydesignsareuniversallyused.Thedifferentstudydesignsacrosseachcategoryaremutuallyexclusivebutnotsowithinacategory.The threeperspectives are thenumberof contacts, the referenceperiodand thenatureof the investigation.The first comprises

cross-sectional studies, before-and-after studies and longitudinal studies. The second categorises the studies as retrospective,prospective and retrospective–prospective. The third perspective classifies studies as experimental, non-experimental and semi-experimentalstudies.Qualitativestudydesignsarenotasspecific,preciseandwelldefinedasdesignsinquantitativeresearch.Also,thereisadegreeof

overlapbetweenstudydesignsandmethodsofdatacollection.Somedesignscaneasilybeconsideredasmethodsofdatacollection.Some of the commonly used designs in qualitative research are: case study design, oral history, focus group studies, participantobservation,communitydiscussionforumsandreflectivejournallog.Fouradditionalapproachestoresearchhavebeendescribed:actionresearch,feministresearch,andparticipatoryandcollaborative

enquiries.Though thesecannot reallybeconsidereddesigns in themselves, theyhaveacquired theirownidentity.Bothactionandfeminist research can be carried out either quantitatively or qualitatively, but participatory and collaborative enquiries are usuallyqualitativeinnature.

ForYoutoThinkAbout

Refamiliariseyourselfwiththekeywordslistedatthebeginningofthischapterandifyouareuncertainaboutthemeaningorapplicationofanyofthemrevisittheseinthechapterbeforemovingon.Identifytwoorthreesituationsrelatingtoyourownareaofinterestwhereyouthinkqualitativestudydesignsmightbemorebeneficialandconsiderwhythismightbethecase.Takeanexamplefromyourownacademicfieldorprofessionalareawhereanexperimental-controlorplacebogroupmightbeusedandexploretheethicalissuesrelatingtothis.