125 ANALYZING THE PERCEPTIONS OF TURKISH UNIVERSITIES USING MULTIDIMENSIONAL SCALING (MDS) ANALYSIS ULAŞ AKKÜÇÜK * SELİN KÜÇÜKKANCABAŞ ** Boğaziçi University Boğaziçi University ABSTRACT With the number of public and private universities in Turkey increasing and with the selection mechanisms offering ever more increased freedom of choice to prospective students, it is becoming important to analyze how students perceive differences among the most preferred Turkish universities. Education marketers may use perceptual maps either to see the current state of the market or to plan for new product launches (in this case, new universities). This paper analyzes the perceptions of the ten most preferred Turkish universities as judged by undergraduate students at universities around Turkey. The Multidimensional Scaling (MDS) technique is used to come up with a perceptual map of the ten most preferred Turkish universities in management education, after obtaining data in the form of similarity judgments, attribute ratings and preference ratings from the respondent students. Additional analyses are performed to attribute meaningful names to perceptual map dimensions, and also to determine the ideal point of the perceptual map. The results indicate that the three private universities plus Galatasaray University form a distinct cluster on their own, while the other six public universities are separated into two distinct groups occupying unique positions in separate quadrants of the perceptual map. It is also found that the ideal point indicates the direction of Boğaziçi University, ODTÜ and İTÜ. Key Words: Turkish universities, perceptual mapping, Multidimensional Scaling, ALSCAL, PROFIT, PREFMAP. TÜRK ÜNİVERSİTELERİNİN ALGILAMALARININ ÇOK BOYUTLU ÖLÇEKLENDİRME (MDS) ANALİZİ İLE İNCELENMESİ ÖZET Giderek artan kamu ve özel üniversite sayısı ve öğrencilerin seçim yapmasında getirilen esneklikler, öğrencilerin üniversiteler arasındaki farkları nasıl gördüklerinin incelenmesini önemli kılmaktadır. Eğitim pazarlamacıları algısal haritaları kullanarak piyasanın mevcut durumunu anlayabilir ya da yeni ürünlerin (yeni üniversitelerin) nasıl konumlandırılması gerektiğine karar verebilirler. Boğaziçi Journal Vol. 21, no. 1-2 (2007), pp. 125-141. * Ulaş Akküçük is an Assistant Professor in the Department of Management at Boğaziçi University, 34342, Bebek, Istanbul, Turkey. E-mail: [email protected]** Selin Küçükkancabaş is a Research Assistant in the Department of Management at Boğaziçi University, 34342, Bebek, Istanbul, Turkey. E-mail: [email protected]
17
Embed
ANALYZING THE PERCEPTIONS OF TURKISH UNIVERSITIES … · ANALYZING THE PERCEPTIONS OF TURKISH UNIVERSITIES USING MULTIDIMENSIONAL SCALING ... of the ten most preferred Turkish universities
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
125
ANALYZING THE PERCEPTIONS OF TURKISH UNIVERSITIES USING MULTIDIMENSIONAL SCALING (MDS) ANALYSIS
ULAŞ AKKÜÇÜK* SELİN KÜÇÜKKANCABAŞ**
Boğaziçi University Boğaziçi University
ABSTRACT
With the number of public and private universities in Turkey increasing and with the selection mechanisms offering ever more increased freedom of choice to prospective students, it is becoming important to analyze how students perceive differences among the most preferred Turkish universities. Education marketers may use perceptual maps either to see the current state of the market or to plan for new product launches (in this case, new universities). This paper analyzes the perceptions of the ten most preferred Turkish universities as judged by undergraduate students at universities around Turkey. The Multidimensional Scaling (MDS) technique is used to come up with a perceptual map of the ten most preferred Turkish universities in management education, after obtaining data in the form of similarity judgments, attribute ratings and preference ratings from the respondent students. Additional analyses are performed to attribute meaningful names to perceptual map dimensions, and also to determine the ideal point of the perceptual map. The results indicate that the three private universities plus Galatasaray University form a distinct cluster on their own, while the other six public universities are separated into two distinct groups occupying unique positions in separate quadrants of the perceptual map. It is also found that the ideal point indicates the direction of Boğaziçi University, ODTÜ and İTÜ.
TÜRK ÜNİVERSİTELERİNİN ALGILAMALARININ ÇOK BOYUTLU ÖLÇEKLENDİRME (MDS) ANALİZİ İLE İNCELENMESİ
ÖZET
Giderek artan kamu ve özel üniversite sayısı ve öğrencilerin seçim yapmasında getirilen esneklikler, öğrencilerin üniversiteler arasındaki farkları nasıl gördüklerinin incelenmesini önemli kılmaktadır. Eğitim pazarlamacıları algısal haritaları kullanarak piyasanın mevcut durumunu anlayabilir ya da yeni ürünlerin (yeni üniversitelerin) nasıl konumlandırılması gerektiğine karar verebilirler.
Boğaziçi Journal Vol. 21, no. 1-2 (2007), pp. 125-141.
Bu makalede işletme eğitiminde en çok tercih edilen on Türk üniversitesinin, Türkiye’de çeşitli üniversitelerde okuyan öğrenciler tarafından nasıl algılandığını çok boyutlu ölçeklendirme (MDS) yöntemi ile incelenmiştir. Öğrencilerden, üniversitelerin birbirine yakınlığı ile ilgili veri toplandıktan sonra, çok boyutlu ölçeklendirme tekniği kullanılarak üniversitelerin algısal haritasına ulaşılmıştır. Farklı bazı teknikler kullanılarak ise MDS sonucu ortaya çıkan boyutlara anlamlı isimler verilmiş ve “ideal” üniversite noktası bulunmuştur. Sonuçlar göstermiştir ki üç özel üniversite ve Galatasaray Üniversitesi birbirine yakın bir küme oluştururken, diğer altı kamu üniversitesi üçerli gruplar halinde farklı yerlerde konumlanmıştır. Ayrıca bulunan ideal noktanın Boğaziçi Üniversitesi, ODTÜ ve İTÜ’nün bulunduğu noktalar yönünde olduğu görülmüştür.
Anahtar kelimeler: Türk üniversiteleri, algısal haritalama, çok boyutlu ölçeklendirme, ALSCAL, PROFIT,
In order for institutions to understand how customers review their products in relation to otherproductsinthemarket,anumberofmultivariatetechniquesfordatavisualizationcanbeused.Thesevisualizationtechniquesgivedecisionmakersasnapshotofhowthecustomersseeproducts(inthiscase,universities)relativetooneanother.MultidimensionalScaling(MDS)isonesuchvisualizationtechnique among other exploratory techniques used to study the interdependence of a number ofvariablessuchasFactorAnalysis,ClusterAnalysisandCorrespondenceAnalysis(Hairetal.,1998).ObtainingperceptualmapsbyMultidimensionalScalingisacommonlyusedmarketingpracticetoshowhowbrandswithinaproductcategoryaresimilartooneanotherandhowtheydifferfromotherbrands(Parasuramanetal.,2004).
InMDS the objective is to convert consumer judgments of similarity (or dissimilarity) betweenobjectsintodistancesrepresentedinmultidimensionalspace(Hairetal.,1998).Thereareanumberofmethodsforhowdatacanbeobtainedorconvertedtoasimilarity(ordissimilarity)measure.Insomecasesdatamaybeoriginallyintheformofsimilaritiesordissimilarities.Agoodexampleisstore-switchingdata(BucklinandLattin,1992).AnotherwidelycitedexampleisMorseCodeconfusiondata(Rothkopf,1957).
Anothermethodwouldbe toobtainsimilarity judgments in the formofasimilarity rating (inourcasewe use a rating scale between one and seven, one corresponding to least similar and sevencorrespondingtomostsimilar)byaskingrespondentstorateallpossiblepairs.Hence,forthethree
Thepaperwillproceedasfollows:Inthenextsection,wewillexplaintheresearchdesignintermsof sampledetermination and characteristics, the selectionof theuniversities to be included in theperceptualmap,theselectionofattributesusedinthePROFITanalysisandthequestionnairedesign.Subsequently,wewillprovidetheresultsintermsofperceptualmapsandcertainnumericalmeasuresfromtheMDSalgorithmALSCALandthecomplementarymethodsPROFIT(toattributemeaningtoMDSdimensions)andPREFMAP(tofindtheidealpoint/vectorinthesameMDSspace).Finally,wewillpresentadiscussionoftheresultsandfuturedirections.
RESEARCHDESIGN
Sample
Oursampleconsistedof594studentsstudyingatdifferentuniversitiesinTurkey.Thesamplewasmainly a convenience sample; however, it wasmade sure that the views of any one university’sstudentsdidnotdominatethesample.Tobemorespecific,405ofthe594studentswerestudentsofthe10universitiesinquestioninthisstudy.ThemaximumnumberofstudentscamefromBoğaziçiUniversitywith13.5%,andthesecondlargestgroupwasİTÜstudentsat13.1%.Alltheothergroupsfell below10%.Also, therewere a total of52universities represented in the sample. In termsofgender,267respondentswerefemaleand327weremale.Intermsofthehighschoolsfromwhichthestudentsweregraduated,apredominantpercentagewereAnatolianHighSchools(322respondents),thesecondlargestwereScienceHighSchools(100ScienceHighSchoolsand14PrivateScienceHighSchools),thethirdlargestwerePrivateHighSchools(89respondents)andthelastwereStateHighSchools(69respondents).
128
Selection of Objects (Universities) to be Plotted on the Perceptual Map
The total number of universities in Turkey is currently 83, comprising 58 public and 25 privateuniversities(YÖK,2007).Askingpairwise judgmentsfromtheentiresetofpairswouldentail therespondentstoanswer3403questions.Thisisobviouslynotfeasibleandmanystudentswouldnotbefamiliarwithavastmajorityofthe83universities.Inordertoreducethenumberofuniversities,weusedtheresultsofthe2005UniversityEntranceExaminationintermsofthenumberofstudentsuniversitiesreceivedfromthetop500.Generally,themoreprestigiousuniversitiesreceivealargernumberfromthe top500.The tenuniversities,BoğaziçiUniversity,HacettepeUniversity,BilkentUniversity,ODTÜ,SabancıUniversity,KoçUniversity,İstanbulUniversity,GaziUniversity,İstanbulTechnicalUniversityandGalatasarayUniversity,altogetherreceived489studentsoutofthe500intheequallyweightedpoints.Wedecidedthatthissmallgroupofuniversitieswouldhavebeeninthehighlycompetitivesetofschoolsfromwhichcompetitivestudentswouldbemakingtheirselections.
Choice of Attributes to be Used in the PROFIT Analysis
Studentsareveryseriousandcarefulwhenchoosingauniversitytoattend.Whentheymakedecisionsabout attending university, and ultimately what university to attend, they consider factors muchdifferentlythanhowpreviousgenerationsdid,suchaseconomic,academic,geographic,cultural,andevenpoliticalfactors(St.Johnetal.,2005;TeachmanandPaasc,1998;WilsonandWilson,1992;Zuker,2006).Ifuniversitiesaretosatisfystudents’requirements,theymustbeawareoftheirownofferingsandhowtheseareperceivedinthemarketplace.Knowingthoseinfluentialfactorsandtheassociatedimpactonpotentialstudentsisimportantforinstitutionalpolicymakers.
Zuker (2006), theVicePresidentandDeanofStudentServicesat theUniversityofDallas,basedon his personal experience and interactionwith the students and their parents, summarized sevenimportantfactorsthathighschoolstudentsshouldconsiderwhenchoosingauniversity.Thesesevenfactorsare thesizeof theuniversity, the locationof theuniversity, theacademicenvironment, thesocial environment, themajors, the extracurricular activities (e.g., drama,debate, journalism, clubsports,studentgovernment),andthecosts.
MazzarolandSoutar(2002)conductedasurveyamong879studentsatAustralianuniversitiesandfoundthatthemostimportantfactorswerethequalityandreputationoftheuniversityandtherecognitionof the institution’s qualifications in their own country. Jackson (1982) noted that studentswouldremovethealternativesonthebasisofgeographic,economicandacademicfactorswiththeevaluationprocessbeingaffectedbyfamilybackgrounds,socialcontextsandacademicexperiences.Chapman(1981)statedthatuniversitychoicewasinfluencedbystudentcharacteristics(socio-economicstatus,aptitude,levelofeducationalexperiencesandhighschoolcapabilities)aswellasexternalmotivations
PreviousworkintheUKconfirmedthatcoursespecifics(content,structure,methodofassessmentofthedegreeprogram)wasthemostpopularlystatedattribute,followedbylocation(distancefromhome,rural/urbanplace,atmosphereofthecampus,facilitiesofthecity/townoftheuniversity)andreputationoftheuniversity(leaguetables,recognizednameordepartment,“old”redbrickuniversities in comparison to “new” universities) (Moogan et al., 1999). For example, somestudentsmightbewillingtochooseauniversitywithanunfavorablelocation(toocloseortoofarfromhome,tooruralortoourban,toobusyortooquiet),inexchangeforamoreappropriatecourse(moreinteresting/relevantmodules,varietyofassessment,inclusionoffieldtrips)whichhadabetterreputation(moreestablisheduniversity) (Mooganetal.,2001).
Gorman(1976)madeadistinctionbetweentheuncontrollablefactorsofhighereducationprovisionsuchaslocationandcontrollablefactorssuchasacademicreputationwherehighstandardscouldbeestablishedandmonitored.Gormanreportedthatlocationandsizewerethecriteriamostfrequentlyused in deciding which university to attend. Reputation for academic quality was of secondaryimportance.
Allofthe45possiblepairsarepresentedhere.ThelowestsimilarityisbetweenKoçandGazi(2.58)and the highest is betweenKoç andSabancı (5.35). In order to get a two-dimensional perceptualmapoftheuniversitiesasseenbythestudents,ALSCAL(YoungandHarris,1990)program(apartofSPSS13.0dataanalysispackage)wasused.SinceALSCALonlyacceptsdissimilaritiesandourdata is originally similarities, the seven-point scalewas reverted such that seven corresponded tomostdissimilarandonecorrespondedtomostsimilar.Afterthisnecessarytransformation,ALSCALwasusedtoobtainatwo-dimensionalsolution.TheresultingmapisgiveninFigure1.Inthismapthefollowingmnemoniccodesareusedinplaceoftheuniversities’fullnames:Bou(Boğaziçi),Bil(Bilkent),Hct (Hacettepe),Odt (ODTÜ),Koc (Koç), Sab (Sabancı), Ist (İstanbulUniversity), Itu(İstanbulTechnicalUniversity),Gz(GaziUniversity),Gs(GalatasarayUniversity).
Figure 1Two-dimensional Perceptual Map of Average Similarities
131
Itisevidentheretheprivateuniversitiesformadistinctgroupatthetopleftcornerofthemap,whilethesixpublicuniversitiesarescatteredacrossdifferentareasofthemap.AlthoughGalatasarayisapublicuniversity, it is locatednear the threeprivateuniversities.Thepublicuniversities formtwogroups:OnegroupinthelowerleftquadrantconsistingofİTÜ,ODTÜandBoğaziçiandtheothergroup located toward therightof thegraphconsistingofGazi,Hacettepeandİstanbul.Hacettepe,whichisactuallyfoundtobeverydissimilarfromallotheruniversities,occupiesadistinctlocationin the lower right quadrant. The closest universities areGazi and İstanbul; howeverHacettepe issomewhatmoredistantthanthegroupcontainingtheGazi-İstanbulpair.
Thenextsectionoutlinesamoreformalapproachinnamingthedimensions,butevenwithoutanymathematical analysis, an examinationof the twodimensionsmay reveal some insights.The firstdimensionmaybeaprivate-publicdimension(fromleft to righton thisdimension:Sabancı,Koç,Bilkent,Boğaziçi,Galatasaray,ODTÜ,İTÜ,Hacettepe, İstanbul,Gazi). Boğaziçi isplacedclosertoprivateuniversitieson thisdimension,perhapsowing to itspast as anAmericanprivatehighereducationinstitution(RobertCollege)inthenearpast.Theseconddimension(fromuptodowninthisdimension:Galatasaray,İstanbul,Sabancı,Koç,Gazi,Bilkent,Hacettepe,Boğaziçi,ODTÜ,İTÜ)maysomewhatreflectthesocialsciencesversusengineeringorientations.Itmayalsorepresentuniversitysize (in termsofnumberofprogramsofferedandnumberof students)and/orhowestablished theuniversityis(withtheupperpositionspertainingtoneweruniversities).Someexceptionsclearlystandouthere,forexample,İstanbulUniversityiswellestablishedandlargebutisontheupperpositions.
Manyinterpretationscanbemaderegardingtheperceptualmapandtheseinterpretationsareonlytheinitialthoughtsoftheauthors.Itisalsoworthwhiletopointoutanyreflection,rotationortranslationofthepointswouldnotchangetheEuclideandistances,andhencewouldgiveessentiallythesamesolution. Interpretationaftersucharotationmaybemoremeaningful.ThePROFITmethod, tobedescribed shortly, does not suffer from this problem as the attributes are plotted as vectors. Therotationwouldnotalterthelocationsoftheuniversitieswithrespecttothevectors.Anothernoteisthatthedimensionalityusedhereistwo,differentdimensionalitiescanalsobeused,andplotscanbegeneratedcomparingpairsofdimensions.threepossibleplotsinthecaseofthreedimensionsandsixpossibleplotsinthecaseoffourdimensions.Thesesolutionsmayalsoleadtodifferentinterpretationsofthedimensions.Thetwo-dimensionalsolutionwillbepresentedhereandlaterthethree-dimensionalsolutionwillalsobepresented.
Thefitofthesolutionisgenerallygood,althoughitcannotbeclassifiedasaperfectorexcellentfit.Thereareanumberofwaysofdetermininghowwellthetwo-dimensionalsolutionsuitstheaveragesimilaritiescalculatedfromthe594subjects(Table1).Thescatterplotofdistancescalculatedfromthe solution given inFigure 1 against the dissimilarities used as input to theALSCALprocedure(generallycalled“disparities,”thesewouldbethevaluesgiveninTable1,exceptthescaleisreversed).ThesmoothnessofthegraphgiveninFigure2indicatesgoodfit.Thissmoothnessisalsocapturedby the square of the correlation between the disparities and the distances. This value, termedR2,indicateshowmuchofthevariationinthedisparitiesisexplainedbythedistancescalculatedfromtheconfigurationfoundbytheMDSprocedure.TheR2valuehereis73.94%,whichisreasonablyhigh.
ThecomputerprogramPROFIT–shortforpropertyfitting–(ChangandCarroll,1989b)ishelpfulin determiningdimensions that are highly correlatedwith the attribute ratings. It employs amoresophisticatedtechniquethantheoneexplainedabove.PROFITtakesthecoordinatevaluesfromtheALSCALoutputand theaverageattribute ratingson thefiveattributesas input.Theoutput is the“directionalcosines”oftheattributes.ThesearepresentedinTable4.Ascanbeseen,thevaluesareveryclosetothecorrelationsreportedinTable3,buttherearesomedifferences.Thehighestvaluesforthefirstdimensionarequalityoffacilitiesandqualityofsociallife.Asfortheseconddimension,
ThecomputerprogramPREFMAP(ChangandCarroll,1989a)takespreferencedataandthestimuluscoordinates obtained from anMDS analysis as input, and provides the “ideal point” in the samecoordinatespaceasoutput.Hence,theso-called“idealuniversity”canbevisualizedintermsofthecoordinatespacealreadygeneratedfortheperceptualmapoftheuniversities.ThisanalysisneedstheALSCALoutputalreadydiscussedandtheaveragepreferencevaluesforeachuniversitycomputedfromtheanswerstoPartCofthesurveyasinput.Theaveragepreferencevaluesfortheuniversitiesaregiven inTable5. It is seenhere thatBoğaziçihas thehighestaveragepreferencewitha largedifference.ODTÜ,Bilkent,İTÜ,Koç,Sabancı,Galatasaray,Hacettepe,İstanbul,andGazifollow.
Figure 3Direction Vectors of Attributes and Universities (PROFIT)
135
ourcaseisfoundtobe(-0.2690,-0.9631).Theplotofthevectorandtheoriginalobjectcoordinatesare provided in Figure 4.Hence, negative direction in the first axis results in a better evaluationofpreferenceandsimilarlynegativedirection in thesecondaxisalsoresults inabetterpreferencerating.Movementinthesecondaxisresultsinagreaterchangeinthepreferencethananequivalentmovementinthefirstaxis.
Higher Dimensional Solutions and Dimensionality Selection
Asindicatedearlier,whiletwo-dimensionalsolutionsaregenerallypreferableandeasytoobtainasanoutputofcomputersoftware,higherdimensionalsolutionscansometimesleadtobetterresults.This is easily portrayed in the plot of the SSTRESSvalues against dimensionality andR2 againstdimensionalitygiveninFigures5and6,respectively.
Figure 4Universities and Ideal Point Vector (PREFMAP)
2), it is readilyevident that this isverysimilar to the two-dimensionalsolutiongiven inFigure1.However,thereseemstobeareflectiononthey-axis.Therelativepositionsoftheuniversitiesremainunchangedandthegroupingsarealsosimilar.Theoriginalinterpretationsoftheaxesstillholdherewiththefirstdimensionrepresentingthepublic-privateandthesecondrepresentingperhapsthesocialscience–engineeringorientations.Thethirddimensionwheninspectedcloselyseemstobeoneoflocation.ThisadditionaldimensionseemstoseparateAnkaraandİstanbul.
Our study (although it has its limitations) shows a number of important properties of BoğaziçiUniversity.Firstofall,theperceptualmapoftheaverageperceptions(Figure1)showsthatBoğaziçi,togetherwithODTÜandİTÜ,occupiesauniquesegment.Uponcloseexaminationwecanalsoseethat Boğaziçi is somewhere in between prestigious public (İTÜ, ODTÜ) and private universities(Sabancı,Koç,Bilkent).Onemayevenstatethatitisa“public”universitythatisofferingmanyoftheadvantagesofa“private”university.Ifwedividetheperceptualmapintofourquadrants,thelowerleftquadrantisoccupiedbyBoğaziçiUniversity,ODTÜandİTÜ.Thisisalsothequadranttowhichalloftheattributevectorspoint.ItisworthwhiletonoteBoğaziçiUniversityhasgottenthehighestscorefromtherespondentsintermsofallfiveattributes.AlsotheidealpointisdirectedatthelocationofBoğaziçiUniversity,ODTÜandİTÜ.
Onemaysaythatsincetheactualuniversityselectiondecisionismadeat thehighschool level, itmaybemore important to study thehigh schoolgraduating class’ opinions inorder toobtain themostaccurateperceptualmap.Afuturestudycanincludehighschoolseniors’perceptions.Anotherrestrictionofthestudyisthelimitednumberofuniversities.Asimilarstudycanbeextendedtoincludemoreuniversities,maybetheentiresetof83universities.However,thelengthofthetimerequiredtofillthesurveysmaybeprohibitivelylarge.Soperhapsasimilarity/dissimilaritymeasurederivedfromattributescanbeused.Todeterminetheattributes,apretestcanbeconducted.
Figure 103-Dimensional Plot D2 against D3
139
Another improvement to the study can be obtained by the use of Individual Differences Scaling(INDSCAL) (Carroll andChang,1970).Thismethod takes inall theproximitymatricesprovidedbytheindividualsandalsofindsouthowindividualsdifferfromoneanotherintermsofhowtheyperceivethedimensions.Thiscanbecomparedtosomeindividualcharacteristicstogainmarketinginsightsintothevariousgroupsofstudents’perceptions.
These programs areMSDOS executables andwere run underMicrosoftXP operating system.2.Ifprospectiveusersareinterested,PREFMAPcanbefoundpackagedwiththetextLattinetal.(2003).ForPROFITrefertoSmith(1989).
REFERENCES
Bucklin,R.E.andLattin.J.M.(1992).“AModelofProductCategoryCompetitionamongGroceryRetailers,”Journal of Retailing,68:271-293.
Chang,J.J.andCarroll,J.D.(1989a).“HowtoUsePREFMAP–AProgramthatRelatesPreferenceDatatoMultidimensionalScalingSolutions,”inP.E.Green,F.J.Carmone,andS.M.Smith(eds.),Multidimensional Scaling: Concepts and Applications:303–317.Newton,MA:AllynandBacon.
------ (1989b). “How toUsePROFIT–AComputerProgramforPropertyFittingbyOptimizingNonlinearorLinearCorrelation,”inP.E.Green,F.J.Carmone,andS.M.Smith(eds.),Multidimensional Scaling: Concepts and Applications:318–331.Newton,MA:AllynandBacon.
Chapman, D.W. (1981). “AModel of Student University Choice,” Journal of Higher Education,52(5):490-505.
Mazzarol, T. and Soutar, G.N. (2002), “‘Push-Pull’ Factors Influencing International StudentDestinationChoice,”The International Journal of Educational Management,16(2):82-90.
Moogan,Y.J.,Baron,S.,andBainbridge,S.(2001),“TimingsandTrade-Offs in theMarketingofHigherEducationCourses:AConjointApproach,”Marketing Intelligence and Planning,19(3):179. Parasuraman, A., Grewal, D., and Krishnan, R. (2004). Marketing Research. Boston: HoughtonMiflin.
Punj.G.N.andStaelin,R.(1978).“TheChoiceProcessforGraduateBusinessSchools,”Journal of Marketing Research,15(4):588.
Rothkopf, E.Z. (1957). “AMeasure of Stimulus Similarity and Errors in Some Paired-AssociateLearningTasks,”Journal of Experimental Psychology,53:94-101.
Russell, M. (2005). “Marketing Education: A Review of Service Quality Perceptions amongInternationalStudents,”International Journal of Contemporary Hospitality Management,17(1):65.
Smith,S.M.(1989).PC-MDS: A Multidimensional Statistics Package.Provo,Utah:BrighamYoungUniversity.
St.John,E.P.,Paulsen,M.B.,andCarter,D.F.(2005).“Diversity,UniversityCosts,andPostsecondaryOpportunity:AnExaminationoftheFinancialNexusbetweenUniversityChoiceandPersistenceforAfricanAmericansandWhites,”The Journal of Higher Education,76(5):545-569.
Teachman, J.D. and Paasch, K. (1998). “The Family and Educational Aspirations,” Journal of Marriage and the Family,(August),60:704-714.
141
Young, F.W. and Harris, D.F. (1990). “Multidimensional Scaling: Procedure ALSCAL,” in J.J.Norusis (ed.),SPSS Base System User’s Guide:396-461.Chicago, IL:SPSSInc.Reprinted inJ.J.Norusis(ed.),SPSS Professional Statistics,(1994):155-222.Chicago,IL:SPSSInc.
Zuker, R.F. (2006) “Factors to Consider in Selecting a University.” Available at: http://www.thehighschoolgraduate.com/editorial/DF/factors[accessedinApril2006].
Wilson, P.M. and Wilson, J.R. (1992). “Environmental Influences on Adolescent EducationalAspirations:ALogisticTransformModel,”Youth and Society,24:52-70.