Manajemen | Fakultas Ekonomi Universitas Maritim Raja Ali Haji joeb.83.3.159-164
Journal of Education for Business
ISSN: 0883-2323 (Print) 1940-3356 (Online) Journal homepage: http://www.tandfonline.com/loi/vjeb20
A Structural Equation Model for Predicting
Business Student Performance
James J. Pomykalski , Paul Dion & James L. Brock
To cite this article: James J. Pomykalski , Paul Dion & James L. Brock (2008) A Structural
Equation Model for Predicting Business Student Performance, Journal of Education for
Business, 83:3, 159-164, DOI: 10.3200/JOEB.83.3.159-164
To link to this article: http://dx.doi.org/10.3200/JOEB.83.3.159-164
Published online: 07 Aug 2010.
Submit your article to this journal
Article views: 51
View related articles
Citing articles: 2 View citing articles
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=vjeb20
Download by: [Universitas Maritim Raja Ali Haji]
Date: 11 January 2016, At: 23:06
Downloaded by [Universitas Maritim Raja Ali Haji] at 23:06 11 January 2016
AStructuralEquationModelforPredicting
BusinessStudentPerformance
JAMESJ.POMYKALSKI
PAULDION
JAMESL.BROCK
SUSQUEHANNAUNIVERSITY
SELINSGROVE,PENNSYLVANIA
ABSTRACT. Inthisstudy,theauthors
developedastructuralequationmodelthat
accountedfor79%ofthevariabilityofa
student’sfinalgradepointaveragebyusing
asamplesizeof147students.Themodelis
basedonstudentgradesin4foundational
businesscourses:introductiontobusiness,
macroeconomics,statistics,andusingdatabases.Educatorsandadministratorscan
useavalidstructuralequationmodelas(a)
acriterionforadmittingstudentstobusinessschool,(b)anadvisingtooltosuggest
tounderperformingstudentsthatamajor
otherthanbusinessshouldbeconsidered,
and(c)atooltosuggestthatmentoringis
necessary.
Keywords:business,gradepointaverage,
macroeconomics,structuralequationmodel
Copyright©2008HeldrefPublications
C
aneducatorspredictthesuccessof
seniorsinanundergraduatebusiness program by early (i.e., freshman
year) performance variables? If valid
performance predictors can be identified, how can the educators apply the
predictors to increase student success
andprogramquality?Thefirstofthese
questions is the primary focus of this
article, in which we describe a structuralequationmodel.Althoughthesecondquestiondeservesmorediscussion,
three applications are readily apparent.
First, using the valid performance predictors, academic advisors can assist
underperformingstudentsearlyintheir
collegecareersinchoosingmajorsoutsidethebusinessschool,ifappropriate.
Second,thesepredictorsidentifyat-risk
studentsformentoring.Third,thesepredictorscouldalsoleadtoanincreasein
student quality because only students
whohaveachievedsomethresholdperformance level would be accepted into
theprogram.
For an Association to Advance CollegiateSchoolsofBusinessInternational
(AACSB)–accredited business program
atasmall,private,liberalartsuniversity,
the success of students (measured by
4-yeargraduationrates,jobplacements,
etc.) is paramount to the mission. The
number of business school students is
restrictedtoaproportionoftotalenrollments to maintain the liberal arts character of the institution. Administrators
strivetofocusresourcesonstudentswho
have the highest probability of success.
Historically, any student could declare
himself or herself a business major on
admission. However, the university is
considering a policy that prevents students from declaring majors until their
sophomoreyear—whichwouldrepresent
a move that could enable the business
school to establish selective admission
criteria. Currently, the business school
requires each student to take the same
foundationcoursesformajorsinbusiness
administration, economics, or accounting. Business majors can choose—duringthejunioryear—anareaofemphasis
withinthebusinessadministrationmajor.
Thecurrentlistofareasofemphasisthat
students can select includes entrepreneurship, finance, global management,
human resource management, informationsystems,andmarketing.
In spring 2006, first-year business
majorstookbusinessfoundationcourses
that included introduction to business,
macroeconomics, statistics, and using
databases. We used students’ grades
from these courses and their Scholastic Assessment Test (SAT) scores,
high school performance, and gender
to construct an initial model to predict
seniorgradepointaverage(GPA).Later
versions of the model eliminated SAT
scores, high school performance, and
gender because these were not statisticallysignificant.
January/February2008
159
Downloaded by [Universitas Maritim Raja Ali Haji] at 23:06 11 January 2016
ReviewofLiterature
A review of business literature
revealedonlyonestudyinwhichcourseworkwasusedtopredictthesuccessof
businessstudents.UsingseniorGPAas
a dependent variable, Brown, McCormick, and Abraham (2002) found that
a single course—principles of macroeconomics—servedasagoodindicator
ofoverallstudentsuccess.Theirmodel
accounted for 36% of the variance in
GPA,andtheyconcludedthatsuccessin
themacroeconomicscoursewashighly
correlated with success in any undergraduatebusinessmajorfield(accounting, economics, finance, information
systems,marketing,ormanagement).
Predictionsbasedsolelyonstudents’
American College Test (ACT) scores
haveshownmixedresults.Brownetal.
(2002) and Laband and Piette (1995)
showedthatACTscoreshadlittletono
impact. However, Naumann, Bandalos,
and Gutkin (2003) found that a student’sACTscorewasasolidpredictor
ofsuccessbutthatusingagroupofselfregulated learning variables (SRLVs)
enhanced the prediction. The SRLVs
include both motivational variables
(intrinsic goal setting, expectancy for
success beliefs, and self-efficacy) and
strategyvariables(studystrategies,goal
setting,andtimemanagement).
BallardandJohnson(2004)studiedthe
linkbetweenstudents’basicmathskills
and their performance in a beginning-
level microeconomics course, which
is typically a more quantitative course
than macroeconomics. Their findings
revealed that the students’ scores on a
faculty-designed,basicmath-skillsquiz
were“positivelyandsignificantlyrelatedtostudentperformance”(p.15).
AsidefromtheworkofBrownetal.
(2002),effortstopredictstudentsuccess
byusingpreviouscourseworkinundergraduate business programs are lacking.Additional studies that incorporate
coursesinadditiontomacroeconomics
forsuchpredictionseempromising.
InitialModelDesign
Weproposedtoconstructapredictive
model of overall student performance,
asgaugedbyseniorGPA,byusingtwo
classesofvariables:studentperformance
and student aptitude. Student perfor160
JournalofEducationforBusiness
mancevariablesarethefinalgradesinthe
four freshman-year foundation courses:
introductiontobusiness(indicatedinthe
AppendixasIBGRD),statistics(SGRD),
macroeconomics(MEGRD),andadatabasecoursethatwascalled“usingdatabases” (UDGRD). These courses are
consistentwiththeAACSBaccreditation
guidelines (2005), and many AACSB
programsmayrequiresimilarcoursesin
eitherthefirstyearorthesecondyearof
theprogram.Aptitudevariablesincluded
verbal SAT score (VSAT), math SAT
score (MSAT), and high school GPA
(HSGPA).Brownetal.(2002)alsoused
severalofthesemeasures.
We included two additional variables—highschoolrank(HSRANK)and
gender (GNDR)—in the initial model.
Highschoolrankisapercentilemeasure
createdastheratioofhighschoolclass
rank to high school class size. This is
the same procedure used by Brown et
al.(2002).Researchershaveusedgender
in previous studies with mixed results.
Brownetal.foundthatgenderwasstatisticallyinsignificant,whereasHolahan,
Green,andKelley(1983),Heath(1989),
and Laband and Piette (1995), all concluded that gender does play a role in
studentperformance.
We collected the data through the
universityregistrar’soffice.Wecollectedgradesofallcurrentbusinessmajors
fromfall2004tospring2004.Themodel
sample included the 147 students who
completed the degree requirements of
theuniversityinMay2004.Thissample
exhibitedaprofileconsistentwiththose
of other recent graduating classes.The
Appendixpresentstheaptitude,performance,anddependentvariablesthatwe
initiallyconsideredinthestudy.
Thisanalysisusedstructuralequation
modeling(SEM),“acomprehensivestatistical approach to testing hypotheses
about relations among observed and
latent variables” (Hoyle, 1995, p. 4).
SEM can “answer a set of interrelated
research questions in a single, systematic, and comprehensive analysis by
modelingtherelationshipsamongmultiple independent and dependent constructssimultaneously”(Gefen,Straub,
&Boudreau,2000,p.4).
The initial structural model, known
as a path model (Hoyle, 1995), examines the effects of each of the aptitude
variables on the grades in each of the
four foundation courses. In addition,
the analysis measures each of the aptitude variables for its direct impact on
theseniorGPA.Also,weexaminedthe
effects of the grades in the foundationcoursesonseniorGPA.Forexample, the independent aptitude variable
VSATmayaffectseniorGPAdirectlyor
through each of the foundation course
grades.Wemeasuredtheimpactofeach
oftheaptitudevariablesonseniorGPA
intwoways:througheachoftheperformancevariablesandonseniorGPA.In
addition,wemeasuredtheimpactofthe
final grade of each foundation course
on senior GPA. We performed all the
analysesaspartoftheSEMapproach.
ResultantModel
WeusedAMOS4.0,aSEMpackage
from SPSS, Inc., to find and confirm
the statistically significant interrelationships that form the resultant path
model. AMOS is a covariance-based
technique.
Figure 1 represents the final resultantpathmodelaftertheanalysisusing
AMOS. The values on the paths are
standardizedcoefficients,andthenumbers above the variables—represented
byboxes—aretheR2values.
Themodelshowsthatthemajorpredictorsofastudent’sseniorGPAarethe
finalgradesthatthestudentreceivesin
thefourfoundationclasses,withmacroeconomicsandstatisticsbeingthemajor
contributors. The student’s final grade
inmacroeconomicswasthekeyindicatorintheBrownetal.(2002)study.
The analysis shows that the model
fitstheobserveddata,asthenormedfit
index(NFI)is0.999,withHoyle(1995)
considering values of 0.9 or greater as
indicating a good fit. In addition, the
model has χ2(3, N = 147) = 3.608, p
= .307. Kline (1998) suggested that a
good fit occurs when the ratio of chisquaretodegreesoffreedomislessthan
3.However,becausethedistributionof
chi-square is such that E(χ2) = df, the
idealfitwouldbewheretheratioofchisquaretodegreeoffreedomisequalto
1(Maruyama,1997).Thus,theNFIand
chi-squaretestsindicatethatthemodel
isagoodrepresentationofthedata,and
themodelfitsthedata.
0.05
Introductionto
business
0.23
0.24
0.29
Downloaded by [Universitas Maritim Raja Ali Haji] at 23:06 11 January 2016
Businessstatistics
0.00
0.18
0.31
0.79
VerbalSAT
0.09
SeniorGPA
0.47
0.33
0.30
Macroeconomics
0.23
0.36
Usingdatabases
FIGURE1.Theresultantpathmodelfromtheanalysisofdata.Thevaluesonthepathsarestandardizedcoefficients,
andthenumberabovethevariable—representedbyboxes—aretheR 2values.
Theresultantpathmodel(seeFigure
1)indicatesthatthegradesinintroduction to business, statistics, and macroeconomics are influenced by the student’s verbal ability—as measured by
theVSATscore.Themodelalsodepicts
the interrelationships among the variables and specifies the significance of
these interrelationships. Table 1 shows
that each variable in the final model,
with the exception of VSAT, is significant at p
ISSN: 0883-2323 (Print) 1940-3356 (Online) Journal homepage: http://www.tandfonline.com/loi/vjeb20
A Structural Equation Model for Predicting
Business Student Performance
James J. Pomykalski , Paul Dion & James L. Brock
To cite this article: James J. Pomykalski , Paul Dion & James L. Brock (2008) A Structural
Equation Model for Predicting Business Student Performance, Journal of Education for
Business, 83:3, 159-164, DOI: 10.3200/JOEB.83.3.159-164
To link to this article: http://dx.doi.org/10.3200/JOEB.83.3.159-164
Published online: 07 Aug 2010.
Submit your article to this journal
Article views: 51
View related articles
Citing articles: 2 View citing articles
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=vjeb20
Download by: [Universitas Maritim Raja Ali Haji]
Date: 11 January 2016, At: 23:06
Downloaded by [Universitas Maritim Raja Ali Haji] at 23:06 11 January 2016
AStructuralEquationModelforPredicting
BusinessStudentPerformance
JAMESJ.POMYKALSKI
PAULDION
JAMESL.BROCK
SUSQUEHANNAUNIVERSITY
SELINSGROVE,PENNSYLVANIA
ABSTRACT. Inthisstudy,theauthors
developedastructuralequationmodelthat
accountedfor79%ofthevariabilityofa
student’sfinalgradepointaveragebyusing
asamplesizeof147students.Themodelis
basedonstudentgradesin4foundational
businesscourses:introductiontobusiness,
macroeconomics,statistics,andusingdatabases.Educatorsandadministratorscan
useavalidstructuralequationmodelas(a)
acriterionforadmittingstudentstobusinessschool,(b)anadvisingtooltosuggest
tounderperformingstudentsthatamajor
otherthanbusinessshouldbeconsidered,
and(c)atooltosuggestthatmentoringis
necessary.
Keywords:business,gradepointaverage,
macroeconomics,structuralequationmodel
Copyright©2008HeldrefPublications
C
aneducatorspredictthesuccessof
seniorsinanundergraduatebusiness program by early (i.e., freshman
year) performance variables? If valid
performance predictors can be identified, how can the educators apply the
predictors to increase student success
andprogramquality?Thefirstofthese
questions is the primary focus of this
article, in which we describe a structuralequationmodel.Althoughthesecondquestiondeservesmorediscussion,
three applications are readily apparent.
First, using the valid performance predictors, academic advisors can assist
underperformingstudentsearlyintheir
collegecareersinchoosingmajorsoutsidethebusinessschool,ifappropriate.
Second,thesepredictorsidentifyat-risk
studentsformentoring.Third,thesepredictorscouldalsoleadtoanincreasein
student quality because only students
whohaveachievedsomethresholdperformance level would be accepted into
theprogram.
For an Association to Advance CollegiateSchoolsofBusinessInternational
(AACSB)–accredited business program
atasmall,private,liberalartsuniversity,
the success of students (measured by
4-yeargraduationrates,jobplacements,
etc.) is paramount to the mission. The
number of business school students is
restrictedtoaproportionoftotalenrollments to maintain the liberal arts character of the institution. Administrators
strivetofocusresourcesonstudentswho
have the highest probability of success.
Historically, any student could declare
himself or herself a business major on
admission. However, the university is
considering a policy that prevents students from declaring majors until their
sophomoreyear—whichwouldrepresent
a move that could enable the business
school to establish selective admission
criteria. Currently, the business school
requires each student to take the same
foundationcoursesformajorsinbusiness
administration, economics, or accounting. Business majors can choose—duringthejunioryear—anareaofemphasis
withinthebusinessadministrationmajor.
Thecurrentlistofareasofemphasisthat
students can select includes entrepreneurship, finance, global management,
human resource management, informationsystems,andmarketing.
In spring 2006, first-year business
majorstookbusinessfoundationcourses
that included introduction to business,
macroeconomics, statistics, and using
databases. We used students’ grades
from these courses and their Scholastic Assessment Test (SAT) scores,
high school performance, and gender
to construct an initial model to predict
seniorgradepointaverage(GPA).Later
versions of the model eliminated SAT
scores, high school performance, and
gender because these were not statisticallysignificant.
January/February2008
159
Downloaded by [Universitas Maritim Raja Ali Haji] at 23:06 11 January 2016
ReviewofLiterature
A review of business literature
revealedonlyonestudyinwhichcourseworkwasusedtopredictthesuccessof
businessstudents.UsingseniorGPAas
a dependent variable, Brown, McCormick, and Abraham (2002) found that
a single course—principles of macroeconomics—servedasagoodindicator
ofoverallstudentsuccess.Theirmodel
accounted for 36% of the variance in
GPA,andtheyconcludedthatsuccessin
themacroeconomicscoursewashighly
correlated with success in any undergraduatebusinessmajorfield(accounting, economics, finance, information
systems,marketing,ormanagement).
Predictionsbasedsolelyonstudents’
American College Test (ACT) scores
haveshownmixedresults.Brownetal.
(2002) and Laband and Piette (1995)
showedthatACTscoreshadlittletono
impact. However, Naumann, Bandalos,
and Gutkin (2003) found that a student’sACTscorewasasolidpredictor
ofsuccessbutthatusingagroupofselfregulated learning variables (SRLVs)
enhanced the prediction. The SRLVs
include both motivational variables
(intrinsic goal setting, expectancy for
success beliefs, and self-efficacy) and
strategyvariables(studystrategies,goal
setting,andtimemanagement).
BallardandJohnson(2004)studiedthe
linkbetweenstudents’basicmathskills
and their performance in a beginning-
level microeconomics course, which
is typically a more quantitative course
than macroeconomics. Their findings
revealed that the students’ scores on a
faculty-designed,basicmath-skillsquiz
were“positivelyandsignificantlyrelatedtostudentperformance”(p.15).
AsidefromtheworkofBrownetal.
(2002),effortstopredictstudentsuccess
byusingpreviouscourseworkinundergraduate business programs are lacking.Additional studies that incorporate
coursesinadditiontomacroeconomics
forsuchpredictionseempromising.
InitialModelDesign
Weproposedtoconstructapredictive
model of overall student performance,
asgaugedbyseniorGPA,byusingtwo
classesofvariables:studentperformance
and student aptitude. Student perfor160
JournalofEducationforBusiness
mancevariablesarethefinalgradesinthe
four freshman-year foundation courses:
introductiontobusiness(indicatedinthe
AppendixasIBGRD),statistics(SGRD),
macroeconomics(MEGRD),andadatabasecoursethatwascalled“usingdatabases” (UDGRD). These courses are
consistentwiththeAACSBaccreditation
guidelines (2005), and many AACSB
programsmayrequiresimilarcoursesin
eitherthefirstyearorthesecondyearof
theprogram.Aptitudevariablesincluded
verbal SAT score (VSAT), math SAT
score (MSAT), and high school GPA
(HSGPA).Brownetal.(2002)alsoused
severalofthesemeasures.
We included two additional variables—highschoolrank(HSRANK)and
gender (GNDR)—in the initial model.
Highschoolrankisapercentilemeasure
createdastheratioofhighschoolclass
rank to high school class size. This is
the same procedure used by Brown et
al.(2002).Researchershaveusedgender
in previous studies with mixed results.
Brownetal.foundthatgenderwasstatisticallyinsignificant,whereasHolahan,
Green,andKelley(1983),Heath(1989),
and Laband and Piette (1995), all concluded that gender does play a role in
studentperformance.
We collected the data through the
universityregistrar’soffice.Wecollectedgradesofallcurrentbusinessmajors
fromfall2004tospring2004.Themodel
sample included the 147 students who
completed the degree requirements of
theuniversityinMay2004.Thissample
exhibitedaprofileconsistentwiththose
of other recent graduating classes.The
Appendixpresentstheaptitude,performance,anddependentvariablesthatwe
initiallyconsideredinthestudy.
Thisanalysisusedstructuralequation
modeling(SEM),“acomprehensivestatistical approach to testing hypotheses
about relations among observed and
latent variables” (Hoyle, 1995, p. 4).
SEM can “answer a set of interrelated
research questions in a single, systematic, and comprehensive analysis by
modelingtherelationshipsamongmultiple independent and dependent constructssimultaneously”(Gefen,Straub,
&Boudreau,2000,p.4).
The initial structural model, known
as a path model (Hoyle, 1995), examines the effects of each of the aptitude
variables on the grades in each of the
four foundation courses. In addition,
the analysis measures each of the aptitude variables for its direct impact on
theseniorGPA.Also,weexaminedthe
effects of the grades in the foundationcoursesonseniorGPA.Forexample, the independent aptitude variable
VSATmayaffectseniorGPAdirectlyor
through each of the foundation course
grades.Wemeasuredtheimpactofeach
oftheaptitudevariablesonseniorGPA
intwoways:througheachoftheperformancevariablesandonseniorGPA.In
addition,wemeasuredtheimpactofthe
final grade of each foundation course
on senior GPA. We performed all the
analysesaspartoftheSEMapproach.
ResultantModel
WeusedAMOS4.0,aSEMpackage
from SPSS, Inc., to find and confirm
the statistically significant interrelationships that form the resultant path
model. AMOS is a covariance-based
technique.
Figure 1 represents the final resultantpathmodelaftertheanalysisusing
AMOS. The values on the paths are
standardizedcoefficients,andthenumbers above the variables—represented
byboxes—aretheR2values.
Themodelshowsthatthemajorpredictorsofastudent’sseniorGPAarethe
finalgradesthatthestudentreceivesin
thefourfoundationclasses,withmacroeconomicsandstatisticsbeingthemajor
contributors. The student’s final grade
inmacroeconomicswasthekeyindicatorintheBrownetal.(2002)study.
The analysis shows that the model
fitstheobserveddata,asthenormedfit
index(NFI)is0.999,withHoyle(1995)
considering values of 0.9 or greater as
indicating a good fit. In addition, the
model has χ2(3, N = 147) = 3.608, p
= .307. Kline (1998) suggested that a
good fit occurs when the ratio of chisquaretodegreesoffreedomislessthan
3.However,becausethedistributionof
chi-square is such that E(χ2) = df, the
idealfitwouldbewheretheratioofchisquaretodegreeoffreedomisequalto
1(Maruyama,1997).Thus,theNFIand
chi-squaretestsindicatethatthemodel
isagoodrepresentationofthedata,and
themodelfitsthedata.
0.05
Introductionto
business
0.23
0.24
0.29
Downloaded by [Universitas Maritim Raja Ali Haji] at 23:06 11 January 2016
Businessstatistics
0.00
0.18
0.31
0.79
VerbalSAT
0.09
SeniorGPA
0.47
0.33
0.30
Macroeconomics
0.23
0.36
Usingdatabases
FIGURE1.Theresultantpathmodelfromtheanalysisofdata.Thevaluesonthepathsarestandardizedcoefficients,
andthenumberabovethevariable—representedbyboxes—aretheR 2values.
Theresultantpathmodel(seeFigure
1)indicatesthatthegradesinintroduction to business, statistics, and macroeconomics are influenced by the student’s verbal ability—as measured by
theVSATscore.Themodelalsodepicts
the interrelationships among the variables and specifies the significance of
these interrelationships. Table 1 shows
that each variable in the final model,
with the exception of VSAT, is significant at p