Accounting Research Center, Booth School of Business, University of Chicago
Educating for Accounting Expertise: A Field Study
Author(s): Dan N. Stone and Marjorie K. Shelley
Source: Journal of Accounting Research, Vol. 35, Studies on Experts and the Application of
Expertise in Accounting, Auditing, and Tax (1997), pp. 35-61
Published by: Wiley on behalf of Accounting Research Center, Booth School of Business,
University of Chicago
Stable URL: http://www.jstor.org/stable/2491452
Accessed: 30-09-2015 17:12 UTC
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.
Wiley and Accounting Research Center, Booth School of Business, University of Chicago are collaborating with
JSTOR to digitize, preserve and extend access to Journal of Accounting Research.
http://www.jstor.org
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
Research
Journalof Accounting
Vol.35 Supplement
1997
Printedin US.A.
Educating for Accounting
Expertise: A Field Study
DAN
N.
STONE
AND
MARJORIE
K. SHELLEY*
1. Introduction
This paper compares the effectivenessof Project Discovery(PD) with
the traditionalaccounting undergraduateprogram at the Universityof
Illinois at Urbana-Champaign (UIUC). PD is a comprehensiveprogram
of instructiondesigned to implementan "educating for expertise"apPD emphasizesacquirproach to accountingeducation. More specifically,
ing intellectualskillsand improvingattitudeswithoutlosing traditional
accounting declarative knowledge. We compare the instructionalprocesses used with,and learningoutcomes of,the firstclass of PD graduates
withsamples of same-semestertraditionalprogramgraduates. Our data
include measuresof course syllabiand students'demographiccharacteristics,declarativeknowledge,intellectualskills,and attitudes.
We analyzedata on instructionalprocesses usingANCOVAsand x2 analysis and data on learning outcomes using ANCOVAsthat include the
*Universityof Illinois at Urbana-Champaign. Sincere thanks to the KPMG Peat MarwickFoundation forgrantsto the firstauthor. Thanks also to AndyCuccia, AudreyGramling, Karen Pincus, Sridhar Ramamoorti,and workshop participantsat the Universityof
Chicago, Southern Illinois University,the Universityof Tennessee, and the Universityof
Illinois for comments on previous drafts.Special thanksalso to the Universityof Illinois
faculty(especially AndrewBailey,Ira Solomon, and David Ziebart) and studentsfor their
cooperation and support. Without their assistance this paper would not exist. Thanks
also to Ross Berman (Coopers and Lybrand), Hamish DeFreitas, Eureka Ellis, Susan
Ambreen Khwaja, and Sabina Zaman (Coopers and Lybrand) for their valuMertzlufft,
able assistance in workingwith the data.
35
1998
of Professional
Accounting,
Copyright
?, Institute
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
36
EXPERTS
AND THE
APPLICATION
OF EXPERTISE:
1997
effectsof instructionalmethod, preaccountingUIUC grade point average (GPA), and ACT comprehensive scores. In addition, because students chose their instructionalprogram,we use a two-stagemodeling
approach describedin Maddala [1983; 1991] to adjust forbias due to students' program choices.1 The resultsare consistentwith the claim that
PD achieved severalof itsgoals withthe firstclass of programgraduates.
Specifically,aftercontrollingforprogramchoice, our resultssuggestthat
PD changed the instructionalprocesses used in the undergraduateprogram, that PD graduates are no worse than traditionalprogramgraduates in theirtraditionalaccounting declarativeknowledge,and thatthey
skillsand attitudestowardaccountancy
have betterproblem-structuring
instructioncompared withtraditionalprogramgraduates. Other results,
while consistentwith the claim that PD achieved its goals, seem more
likelyto resultfromstudents'programchoices ratherthan instructional
method.
Our workcontributesto researchon knowledge,learning,and expertisein accountingdomains in at least threeways.First,we studyan actual
accountingeducation change initiative.This approach rules out "systematic" (i.e., factorial,orthogonal) research designs which can sometimes
provide unequivocal casual inferenceslinking experimentalmanipulations withoutcomes.2 Our approach instead is to capture the actual relationshipsamong variables in a naturallyoccurring setting (Brunswik
[1956], Hammond [1966], and Libby [1981]).
Our design thereforeproduces potentiallymore generalizable results.
But this generalizabilitycomes at a cost: we cannot specifythe causal
mechanismsthat are responsible for the performanceof PD graduates.
Therefore,while we provide evidence on the relative effectivenessof
PD, we cannot point to the treatmentor treatmentsthat produce these
state whetherthe observed outeffects;further,we cannot definitively
comes resultfrominstructionalmethod differencesor fromstudentcharacteristicsthatled to theirprogramchoices.
A second contributionof our studyis thatwe develop and adapt measures of knowledge,skills,and attitudesthat are specificto accounting.
Previousresearch in cognitivepsychology(Gardner [1983; 1991; 1993],
I Various termsare used in the research literatureto describe the effecton estimation
and inferenceof human choices thatinfluenceoutcomes related to those choices. For example, Maddala [1983; 1991] and some accounting researchers(Yeo and Ziebart [1995a;
Achen [1986] and others (e.g., Cook and
1995b]) referto such choices as "self-selection."
Campbell [1979]) discussself-selectionas one possible uncontrolledfactorin quasi-experiments in which researchers cannot assign participantsto conditions. We use the term
"programchoice" to referto students'choices of the PD or traditionalprogram.
2 An alternativeexperimentaldesign would have been to assign participantsrandomly
to the PD and traditionalprograms. But even if such assignmentswere possible, they
would be undesirable. In field research,random assignmentof participantsto conditions
oftendecreases internalvalidityby producing unintended reactionsto experimentaltreatments (e.g., see Blumbergand Pringle [1983], Cook and Campbell [1979], and Koch and
Rhodes [1979]).
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
EDUCATING
FOR ACCOUNTING
EXPERTISE
37
Frensch and Sternberg [1989], and Schneider [1989]) and auditing
(Libby and Tan [1994, p. 710]; see also Marchant [1990]) indicates that
knowledgeand skill are domain- and sometimeseven task-specific.The
localized, contingentnature of expertisemakes it crucial that accounting researchassess knowledgeand skillin accounting tasksand contexts
rather than using generalized, context-independentpersonalitymeasures (e.g., cognitivestyle).With one exception (tolerance for ambiguity), our measures of post-programknowledge,skills,and attitudesare
specificto accounting tasksand contexts.
Third, we present field evidence that cognitive learning principles
can guide instructionalprocesses in a large-scale curriculumrevision.
Existingexperiments(e.g., Bonner and Walker [1994]) demonstrateapproaches for applyingthe insightsof cognitiveresearch to the problem
of developing expertisein accounting settings.But we know of no field
demonstrationsof the efficacyof such approaches to learning accounting knowledgeand skills.While we do not presenta comprehensivecostbenefitanalysisof PD, we do presentevidence suggestingthat cognitive
learning principlescan guide undergraduateinstructionin accounting.
In addition, our futureworkincludes assessmentof the program'slongtermoutcomes.
In the followingsections,we describe PD's goals, the researchhypotheses, the methodsused to testthe hypotheses,the results,and the implications and limitationsof our study.
2. EducatingforAccountingExpertise
Categoriesof LearningOutcomes. Cognitive research provides insight
into the processes by which novices become experts (Bedard [1989;
1991], Bonner and Pennington[1991], Chi, Glaser,and Farr [1988], Glaser [1989], and Libby [1995]). This researchsuggestsdifferingcategories
of learningoutcomes thatrequire differentinstructionalapproaches and
make differentinformation-processing
demands forlearning (Anderson
[1985], Gagne [1984], and Winograd [1975]). Three importantcategories of learning outcomes in educating for expertise are declarative
knowledge,intellectualskills,and attitudes.3
Historically,accounting education has emphasized declarativeknowland vocabulary)(Bonner and Walker[1994]).
edge (i.e., facts,definitions,
Intellectualskills(also called procedural knowledge) include knowledge
3As manyauthorsdiscuss (e.g., Bonner and Walker [1994] and West,Farmer,and Wolff
[1991]), declarativeknowledge and intellectualskills are neither mutuallyexclusive nor
independent since intellectualskillsoftenrequire declarativeknowledge. Despite thisimprecision,these are the fundamentallearning outcomes generallyidentifiedin psychological theoriesof expertise and expert knowledge (e.g., Anderson [1976; 1985] and Gagne
[1984]). In addition, these outcomes roughlycorrespond to those identifiedby the Accounting Education Change Commission [1990] as relevantto the futureof accounting
education.
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
38
DAN N. STONE
AND
MARJORIE
K. SHELLEY
of the concepts,rules, and procedures needed for problem solvingin a
domain (Anderson [1976; 1985]). Withinaccounting, intellectualskills
include the ability to differentiateand integrate alternativeproblem
perspectives,the abilityto identifyaccounting-relatedinformationresources, the abilityto structureproblem solutions,and writtencommunication skills (e.g., see Baker et al. [1995], Big Eight AccountingFirms
[1989], Deppe et al. [1991], and Francis, Mulder, and Stark [1995]).
Attitudesare beliefsthatinfluenceindividuals' choices and actions (Allport [1935] and Gagne [1984]). Attitudesdeterminethe extentof recognition thatlearning is needed (Feldman [1986]) and stronglyinfluence
the abilityto recognize the need for and apply intellectualskills (Ennis
[1987]).
The goal of the PD curriculumis to change the instructionalprocesses
at the UIUC to an "educating for expertise"approach emphasizing the
developmentof intellectualskillsand attitudes.4It is thusa comprehensive revisionof both the formand contentof the undergraduatecurriculum, based on the contractingrole of accounting-relatedinformation
(Solomon [1997]). The PD facultyand administratorsredesigned every
undergraduateaccounting course by applyingand adapting activelearning and educating forexpertiseprinciplesto accounting education. The
redesigned curriculumincludes a two-coursesequence that provides a
conceptual introductionto accounting; a five-course"core" sequence
structuredaround functionalbusiness concepts such as decision making
and measurement;an evolving set of courses designed to develop intellectual skillsand professionalattitudes;and elective courses that introduce students to the professional rules and standards of auditing,
financialreporting,and taxation.
The PD curriculum applies methods from research on developing
expertisein professionaldomains. These methods include:
and casessimiill-structured,
ambiguousproblems
(1) Focusingon complex,
lar tothose
practice.Research indicatesthatdeveloping
foundin accounting
professionalexpertiserequires analyzingthe issues and settingsthatare
central to professionalpractice (Gardner [1993]). These issues and setcomplex and ambiguous to providefutureprotingsmustbe sufficiently
fessionalswith "simulated" experience in the domain where theywill
practice.At the same time,instructionmuststructureand organize these
experiences to make learning possible (Bartunek,Gordon, and Weathersby[1983]). The PD facultydeveloped cases and problemsthroughout
the curriculumto achieve these learning outcomes.
4 The goal of the PD programis not to create accounting expertsbut ratherto provide
graduateswiththe skillsessential to developing expertise.This goal is consistentwiththe
AECC's perspective:"Atthe time of entry,graduates cannot be expected to have the range
of knowledgeand skillsof experienced accountingprofessionalaccountants.To attainand
maintain the statusof a professionalaccountant requires continual learning. Therefore,
pre-entryeducation should lay the base on which life-longlearning can be built" (AccountingEducation Change Commission [1990, p. 305]).
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
EDUCATING
FOR ACCOUNTING
EXPERTISE
39
(2) Developingintellectualskillsusing activelearningapproaches.Active
learningis a core instructionalapproach of the PD curriculum(Bonwell
and Eison [1991]) that develops intellectualskillsby teaching students
to identifyand integratemultiple perspectiveson complex accounting
problems into well-reasoned,defensibleanalyses.Studentsalso learn to
workin teams and to structureand present oral and writtenarguments
(Francis,Mulder, and Stark [1995]).
(3) Repeatingthemes,materials,and cases. Repetition transformsthe
"controlled"processing of novices into the "automated" processing of
experts (Gardner [1993] and Schneider [1989]). OrganizingPD around
the contractingrole of accounting provides a centralfocus for organizing within-and between-classinstructionand for creatingthe necessary
repetitionof essential skills (UIUC, Departmentof Accountancy[1995]
and Solomon [1997]).
thatpromoteslearningand innovation
(4) Creatinga social environment
(Feldman [1986]). PD includes effortsto develop a self-directedlearning environmentthat extends beyond the classroom. These effortsinclude regular formal and informal interactions between practicing
accounting professionalsand PD students,a PD students' association,
and severalfaculty/student
events (e.g., faculty/student
lunches, "open"
officehours, facultyattendance at PD studentgatherings).
(5) Emphasizingthe development
of intellectualskills and attitudesand
declarativeknowledge(cf. Feldman [1986] and Gardner
de-emphasizing
[1993]). Traditionally,accounting education has focused heavily (and
some argue inappropriately) on developing declarative knowledge
(AECC Position StatementNo. 1, Big Eight Accounting Firms [1989]).
In contrast,cognitivelearning approaches often emphasize developing
the intellectual skills associated with domain-specificproblem solving
(Anderson [1985]). Consistentwith cognitivelearning approaches, the
PD curriculumemphasizes developing accounting-relevantintellectual
skillsand attitudes.
PD facultyare encouraged to offer
(6) Changingextrinsic
incentives.
extrinsicincentives (e.g., class credit) for performanceon ambiguous,
complex cases and fieldprojectsand to de-emphasize declarativeknowledge exercises such as structuredproblem examinationsand quizzes. In
addition, the PD facultymet to discuss and implementlearning activities that rewarded insightfulanalyses and justifiable decision processes
and conclusions, not simply"correct"solutions.
3. Programmatic
Goals and ResearchHypotheses
3.1
HYPOTHESIS-INSTRUCTIONAL
PROCESSES
To provide evidence on the extent to which PD affectedinstructional
methods,we interviewedPD and traditionalprogram instructors,compared PD and traditionalclass syllabito determinethe extrinsicincentives
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
40
DAN N. STONE
AND
MARJORIE
K. SHELLEY
offeredfordifferingtypesof assignments,and compared PD and traditional graduates' responses to questions about instructionalprocesses.
HI:
3.2
The ProjectDiscoveryprogramincludes more expertise-building
instructionalprocesses than the traditionalprogram.
HYPOTHESIS-DECLARATIVE
KNOWLEDGE
The PD curriculumde-emphasizes the directteaching of declarative
knowledge in favor of developing intellectual skills and attitudes. PD
acquire declaraalso seeks to improvestudents'abilitiesto independently
tiveknowledge. One possible outcome is thatPD graduates' declarative
knowledgemightdecrease relativeto traditionalprogramgraduates.Albecause of PD's emphasison independentlyacquiring declarternatively,
ative knowledge,PD graduates mighthave more declarativeknowledge
than traditionalprogramgraduates. Based on this discussion,we state a
null prediction:
H2: There will be no differencein declarative knowledge among
Project Discoveryversustraditionalprogramgraduates.
3.3
HYPOTHESIS-INTELLECTUAL
SKILLS
The PD aims to improveintellectualskillswithinaccountancy (Deppe
etal. [1991]).
is the abilityto differentiatealternativeper(1) Cognitivecomplexity
spectivesand integratethese perspectivesinto a well-reasoned analysis
(Streufertand Swezey [1986]). Cognitivecomplexityis one cornerstone
of criticalthinking(Ennis [1987] and Francis,Muldar,and Stark[1995]).
In addition, increasing cognitive complexityis an importantlearning
objectiveof instructionalstrategiesbased on cognitiveresearch (Breuer
and Tennyson[1995]). These argumentssuggestthatdeveloping and assessingcognitivecomplexityis importantforimprovingproblem-solving
skillsin complex,ambiguousenvironmentssuch as accounting.Bartunek,
Gordon,and Weathersby[1983] suggesteducational methodsforincreasing cognitivecomplexity,many of which were implemented in the PD
program (e.g., analyzingproblems frommultipleperspectives,dialectic
inquiry).
Identifyresources.
information
accounting-related
(2) Abilityto identify
is
resources
information
accounting-related
ing and using appropriate
learning
self-directed
[1987]),
(Ennis
thinking
to
critical
fundamental
(Francis,Mulder, and Stark [1995]), and expertise (Bedard [1989]). PD
includes instructionspecificallydesigned to develop students' abilities
to independentlyacquire and use the informationneeded for solving
and using on-line and traditional
accounting problems (e.g., identifying
data sources).
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
EDUCATING
FOR ACCOUNTING
EXPERTISE
41
skills.Problem
and (4) Writtencommunication
(3) Problemstructuring
structuringis the abilityto organize and structuresolutions to ambiguproblems. Many authors identifyproblem-structuring
ous, ill-structured
and writtencommunicationskillsas criticalcapabilities for accounting
professionals (e.g., Deppe et al. [1991], Francis, Muldar, and Stark
[1995], and Gainen [1992]). The PD programprovides extensivetrainand
ing in and supportforimprovingparticipants'problem-structuring
writingskills.5
In summary,we predict PD graduates will have higher intellectual
skills measures in accounting-relatedtasks relative to traditionalprogram graduates:
H3:
3.4
Project Discoverygraduates will have higher levels of accounting-relevantintellectual skills relative to traditional program
graduates.
HYPOTHESIS-ATTITUDES,
TOLERANCE
FOR AMBIGUITY
Changes in accounting practice make accounting professionals'tolerance for ambiguityimportantfor success (Chen and Olsen [1989] and
Pincus [1991] ). Tolerance forambiguityalso affectsthejudgmentsof professionalaccountants;forexample, Pincus [1990] reportsthatambiguitytolerant auditors are more confidentin their fairnessof presentation
judgments of financialreports,and Gupta and Fogarty[1993] find that
auditors' tolerance for ambiguityshaped theirperceptions of the audit
inprocess. Because previous evidence suggeststhat ambiguity-tolerant
PD
includes
situations,
ambiguous
can
better
complex,
manage
dividuals
instruction(e.g., activelearningmethods) thatshould increasegraduates'
tolerance forambiguity:
H4a: Project Discoverygraduateswillhave highertolerance forambiguitythan will traditionalprogramgraduates.
3.5
HYPOTHESIS-ATTITUDES,
ATTITUDES
TOWARD INSTRUCTION
Universityprogramgraduatesare importantstakeholderswhose opinions partiallydetermine programmaticsuccess (Alkin and Associates
[1985] and Stufflebeamand Shinkfield[1]985]). Increasingly,both public
and private accounting programs depend on alumni contributionsto
support their research and teaching missions. For these reasons, program graduates' attitudestowardinstructionare importantin evaluating
5For example, the Department of Accountancyhired a facultymember with English
and writtencommunication asand Accounting degrees to develop problem-structuring
signmentsand to support PD's learning outcomes related to these skills.
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
42
DAN N. STONE AND MARJORIE K. SHELLEY
1
TABLE
and Measures
Goals,ResearchHypotheses,
Programmatic
ProjectDiscovery
ProgrammaticGoals and Hypotheses
Measures
Panel A: InstructionalProcesses
HI (manipulation check): The Project Discoveryprogramuses more expertisebuilding instructionalprocesses than the traditional
program.
Increase the importanceof field
projectsand cases, group work,
and presentationsin accountancy
departmentclasses
1. Course instructorinterviews
2. Point allocations on course syllabi
3. Surveysassessing studentperceptions
of instructionalprocesses
Panel B: Declarative Knowledge
H2: There will be no differencein declarativeknowledge among Project
Discoveryversustraditionalprogramgraduates.
No decrease in declarativeaccounting
knowledge
1. Contentanalysisof essaysfortechnical
accounting knowledge
2. CPA examination performance
Panel C: IntellectualSkills
H3: The Project Discoverygraduateswill have higher levels of accountingrelevantintellectualskillsrelativeto traditionalprogramgraduates.
1. Increase cognitivecomplexity
2. Increase abilityto identifyaccountingrelated informationresources
3. Increase abilityto structureproblem
solutions
4. Increase writtencommunicationskills
1. Content analysisforcognitive
complexity
2. Content analysisforidentification
of accounting-relatedinformation
resources
3. Holistic evaluation of problem
structuring
4. Holistic evaluation of writingquality
Panel D: Attitudes
H4a: Project Discoverygraduateswill have higher tolerance forambiguity
than will traditionalprogramgraduates.
H4b: Project Discoverygraduates will rate accountancydepartmentinstruction
more highlythan will traditionalprogramgraduates.
1. Increase tolerance forambiguity
2. Improve attitudestowardaccountancy
instruction
3. Improvebeliefsabout the extent
to whichaccountingcourses have
increased graduates' knowledge
and skill
1. AT-20 measure of tolerance for
ambiguity
2. Student perceptions of instructional
processes
3. Student perceptions of effectsof
courses on knowledge and skill
programmatic success (Demong, Lindgren, and Perry [1994]). PD
sought to increase graduates' perceptions of the qualityof accountancy
instruction:
H4b: Project Discoverygraduates will rate accountancy department
instructionmore highlythanwilltraditionalprogramgraduates.
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
EDUCATING
FOR ACCOUNTING
EXPERTISE
43
Table 1 summarizesthese hypothesesand the related program goals
and measures.
4. Methods
4.1
RESEARCH
DESIGN
and Participants. The research design is a two-group
4.1.1. Overview
comparison (i.e., PD versusthe traditionalprogram) thatincludes measures of potentiallynonequivalent concomitant variables and models
participants'choices of instructionalprograms.We collected data comparing the instructionalprocesses in the PD and traditionalprograms
and the declarative knowledge, intellectual skills,and attitudesof the
firstclass of graduates of the PD programwithgraduatesfromthe same
semesterof the traditionalprogram.We also gathered data on graduates' demographic characteristicsto assess theirpotential effecton participants'choices of programsand programoutcomes.
Ethical and universityrequirementsprevented us fromrandomlyassigningstudentsto (i.e., forcingstudentsto enroll in) the PD and traditional programs. Participantschose the PD program in response to
threeinformationalmeetingsheld in the springprior to the semesterin
stuwhich the firstPD courses were offered.One hundred eighty-five
dents applied foradmission to the PD program.Applicantswho were on
academic probation,had alreadytaken introductoryaccounting,or had
not been admittedto the College of Business Administrationcould not
enroll in PD.
Afterscreening, 164 students enrolled in, and 148 completed, the
introductoryPD course. Three years later, 83 graduates
first-semester
completed the PD program.6During the same semester,279 graduates
completed the traditionalprogram. Except for demographic measures,
we report data on randomlychosen PD graduates and traditionalprogram graduates fromrandomlysampled courses. Sample sizes range between 60 and 83 for PD graduates and between 86 and 205 for
traditionalprogramgraduates.
4.1.2. Procedures.We distributedassessment-relatedquestionnaires,
cases, and essay questions to either a random sample (for the questionnaires and essayquestions) or a census (forthe cases) of PD studentsand
to random samples of traditionalprogramcourses in the fall and spring
semestersof students'senioryears.Studentsreceivedbetween1% and 3%
6 Sixty-four
studentsenrolled in and completed part of the PD programbut graduated
in the traditionalprogram.We reanalyzed the data withthese "partial"Project Discovery
students (PPD) as a separate group. The resultsindicate that the PPD studentswere academicallyweakerand less satisfiedwithDepartmentof Accountancyinstructionthan either
the PD or traditionalprogram graduates. A summaryof these analyses is available upon
request.
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
44
DAN N. STONE
AND
MARJORIE
K. SHELLEY
and carefully"completingthe asof theircourse grade for "thoughtfully
sessmentexercises.7
4.2
MEASURES
Processes. We tested for differencesin instruc4.2.1. Instructional
tional processesbetween the core PD and traditionalprogramcourses by
interviewingthe 9 core-course PD instructorsand 9 of the 13 traditional program instructorswho taught identicallynumbered courses.
These interviewsindicated thatPD courses used more small-groupwork,
in-class discussion, case and field projects, and student presentations,
while traditionalprogram courses used more instructorlectures,structuredhomeworkproblems,examinations,and quizzes.
To confirmthese qualitativedata and provide evidence on the extrinsic incentivesofferedin the PD and traditionalprograms,we obtained
and compared syllabifor the introductoryand core PD courses (n = 7)
and matched traditionalcourses (n = 14). In addition, we surveyed62
PD graduates' and 153 traditionalprogramgraduates' perceptionsof instructionalprocesses.
4.2.2. ParticipantDemographics.We obtained confidential data on
ethnicity,gender, number of high school accounting classes completed,
and ACTscoresfor69 PD graduatesand 188 traditionalprogramgraduates.
4.2.3. Declarative Knowledge-CPA Examination Performance.Although performanceon the CPA examination probably requires both
declarativeknowledgeand intellectualskill,we categorizedparticipants'
CPAexaminationperformanceas a measure of declarativeknowledge.We
obtained releases allowingus access to the CPAexaminationscores of all
PD (n = 70) and traditionalprogram (n = 130) graduateswho indicated
theyplanned to take the CPA examination in the next year. The State
Board of Accountancyprovidedthese scores.
4.2.4. Intellectual
Skills-Case Analysis. Three weeks beforethe end of
the fall semesterof theirsenior year,participantsanalyzed a case which
asked themwhat theywould do, and where theywould seek information,
in response to a CEO who was reluctant to disclose informationabout
a contingentliability.We used Suedfeld, Tetlock,and Streufert's[1992]
protocol to measure the cognitivecomplexityof participants'analysisof
these cases. This protocol includes procedures and practice cases for
trainingcoders and forcoding and analyzingtextsforcognitivecomplexity. We trained two coders in this protocol and tested that they had
achieved acceptable skilllevels.The coders then analyzedeach paragraph
of participants'essay responses for cognitive complexity.Participants'
overall complexityscores were the average of the complexityscores of
the paragraphsin theiressays.Both coders analyzed a randomlychosen
7We returnedincomplete and frivolousresponses (less than 5% of the responses from
any group) to studentsfor completion.
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
EDUCATING
FOR ACCOUNTING
EXPERTISE
45
sample of 50 essays.The correlationbetweencoders cognitivecomplexity
codingswas .90 (Cohen's Kappa = .64, Cohen's Kappa standarddeviation
= .07). The coder withmore accounting experience then coded the remainingessays.
Participants' responses were also content analyzed for intellectual
skillsin identifyingaccounting-relatedinformationresources,problem
structuring,and writing.8Coders made these assessmentson 10-point
scales rangingfromverypoor (1) to excellent (10). Aftertrainingin the
content analysisprotocol, two coders analyzed a random sample of 30
essayresponses. The correlationsbetween coders ranged from.84 to .95
for these categories.The coder withmore accounting experience completed content analysiscoding.
4.2.5. Attitudes.Three weeks prior to graduating,participantscompleted instrumentsmeasuring their tolerance for ambiguityand attitudes toward accountancy instruction.Consistent with some previous
researchin accounting (e.g., Pincus [1990; 1991]) we measured tolerance
for ambiguityusing the AT-20 scale (MacDonald [1970]). Higher AT-20
scores indicate higher tolerance for ambiguity.Tests of internalconsistencysuggestlevels similarto those found in previous research (Cronbach's a = .62 [n = 202] vs. .64 reportedin MacDonald [1970]).
We assessed graduates' attitudestoward accountancyinstructionwith
six surveyquestions administeredthreeweeks prior to graduation.Cronbach's a statisticindicated high internal consistencyin these measures
(a = .85).
4.3
METHODS
OF ANALYSIS
We used t-teststo examine Hypothesis1 data related to syllabipoint
allocations for the PD and traditional programs. We used chi-square
analyses to test for differencesbetween the PD and traditional programs in participants' categorical surveyresponses. For all intellectual
skilland attitudemeasures,we testedforthe effectsof instructionalprogram (i.e., PD or traditional) using an ANCOVAmodel that included
instructionalprogram, preaccounting UIUC GPA, and ACT comprehensive score, as well as a composite variable adjustmentfor program
selection bias. For the attitude analyses, a measure of attitude toward
personal competence was included. To investigateCPAexaminationdata,
we added graduates' CPA examination review course type and grade
(i.e., UIUC course grade = "A,"UIUC course grade = "B,"not enrolled in
UIUC reviewcourse) as an explanatoryvariable in the above-described
ANCOVAmodel.9
8We used protocols developed by the Educational TestingService for the holisticgrading of writtencommunication to assess writingquality (Educational Testing Service
[1987], Fowles [1990], and GRE FacultyAdvisoryTerm [1991]).
9We used one-tailed hypothesistestsfordirectionalpredictionsand two-tailedhypothesis testsforall other explanatoryvariables in the models.
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
46
DAN N. STONE
AND
MARJORIE
K. SHELLEY
4.3.1. Analysisof ProgramChoice. When treatmentgroups differon
characteristicsthat covary with outcome measures, analyses that omit
those characteristicsproduce biased estimatesof the treatmenteffect
(Achen [1986] and Maddala [1983; 1991]). While the directionof the
bias depends on the correlationsbetween the omitted characteristics,
group membership,and the dependent variable, in general, the bias
favorsfindingtreatmenteffects.Omittingselection characteristicscan,
of PD.
therefore,lead to mistakeninferencesabout the effectiveness
While statisticiansdisagree on the best approach to controllingfor
potentialbiases resultingfromhuman choices, mostargue forthe use of
one of two possible approaches (e.g., see Achen [1986]). In the singlestage approach, the characteristicshypothesizedto explain choice are
included as explanatoryvariables in an ANCOVAor regressionmodel,
along withthe treatmentvariable. This approach can, in a single model,
capture the effectsof both the programand the skillsand aptitudesthat
influencedprogramchoice. One limitationof thisapproach is thateach
added measure contains errorand, in general, the effecton estimatesof
the error in this settingis the same as the effectof selection bias; it
favorsfindinga treatmenteffect(Achen [1986]). A two-stageapproach
reduces the effectof errorin the explanatoryvariables (Achen [1986]).
In a two-stage approach, the characteristicsleading to program
choice are used to forma single compositevariable thatreplaces the set
of measures used to control for self-selectionin the single-stageapproach. For example, Maddala [1983; 1991] argues for the use of maximum likelihood probit analysis to model individual choices.10 The
compositevariable used in our analysesis the inverseMills ratio,a transformationof the probit analysisfittedvalues. This composite variable
has been "purged" of the error contained in the individual measures
used as explanatoryvariables in the probitmodel (Achen [1986]).
4.3.2. AnalysisofProgramChoice-ModelofProgramChoice. We chose
variablesforthe model of participants'programchoices based on speculation regardingvariables that seemed likely to predict programmatic
and suggestionsmade at theJournalofAccounting
choices, data availability,
ResearchConference.11Including these variablesmateriallydecreases the
class of graduates,
sample sizes available forthe analysis;forthe first-year
the total sample decreases fromapproximately250 to 105. In addition,
the decrease is asymmetric;the data for80.72% of the PD graduatesare
available for the model of programchoice, while the data for 22.16% of
the traditionalprogramgraduatesare available.
10See Shehata [1991] and Yeo and Ziebart [1995a; 1995b] for previous accounting applications of this approach.
11Based on suggestionsmade at theJournalofAccounting
ResearchConference,we also
requested data from universitysources on graduates' specific-courseperformance.We
hoped to use these data to improve the predictivepower of the program choice model.
Unfortunately,
these data were unavailable.
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
EDUCATING
FOR ACCOUNTING
EXPERTISE
47
of the data
To increase the statisticalpower of testsand the symmetry
set, we included data for studentsfromboth the firstand second years
of the PD and related years of the traditionalprogramsin the data set
used to fitthe programselection model. The resultingsample (both the
firstand second years' data) contains 268 PD students and 231 traditional programstudents.12
Unfortunately,
some of the data were collected at differenttimesfor
the year-oneand year-twostudents.Specifically,we firstcollected several
of the measuresin the programchoice model afterthe year-onestudents
had completed one year of college accounting but at the beginning of
the year-twostudents'firstcollege accounting course. Because of these
timingdifferences,we tested for parameter estimatedifferencesin the
models of the first-and second-yearstudents' data using a Chow test
(Kmenta [1986]). The Chow testresultsled us to create separate explanand second-yearstudents' responses to
atoryvariables for the first-year
the statement,"accountingis interesting"(yes = 1, no = 0), and to add a
term that measured the change in the effectof general self-efficacy
(GSE) influenceon year-twostudents'programchoices.
The resultingprobitmodel of students'programchoices is:
Yi=
++ hXI +
+ 53X3 + f4X4 + 35X5+ 56X6 = :7X7 + J8X8, (1)
where:
Vi = program choice (PD = 1, tradition= 0),
XI = year-one students' response to "accounting is interesting"
(yes = 1, no = 0).
X2 = year-twostudents' response to "accounting is interesting"(yes
= 1, no = 0).
X3 = responses to "accounting is a highlystructuredsubject" (yes =
1, no = 0).
X4 = responses to "did you take high school accounting?" (yes = 1,
no = 0).
X5 = preaccounting UIUC grade point average,
X6 = gender (male = 1, female = 0),
X7 = general self-efficacy,13
foryear-twostudents.
X8 = general self-efficacy
12The sample includes 67 year-onePD students,37 year-one traditionalstudents,201
year-twoPD students,and 194 year-twotraditionalstudents.
13 General self-efficacy
(GSE) measures one's beliefs regardingwhetherone is, in general, a competent,effectiveindividual (Sherer et al. [1982] and Tipton and Worthington
[1984]). GSEjudgments are hypothesizedto resultfromprevious experiences, which create expectations about futuresuccess in related or new situations.GSE is hypothesizedto
be an enduringindividualtraitthatis resistantto short-termexperiences of success or failure (Eden and Kinnar [1991]). For the 499 subjects in the program choice model, the
reliability(as measured using Cronbach's at) of the GSE measure is .91.
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
DAN N. STONE AND MARJORIE K. SHELLEY
48
TABLE
2
ProcessResults
Instructional
Panel A: PointAllocations on PD and TraditionalProgramSyllabi
Traditional
PD Program
Program
Percentage of Class
(n = 14)
t-Statistic*
(n = 7)
PointsAllocated to
Field Research and Cases
StructuredProblem
Examinations and Quizzes
Other Activities
p
38.4%
4.6%
6.0
<.001
55.1%
6.4%
89.9%
5.5%
5.7
0.2
<.001
.848
100%
99.9%
Totals
equalityofvariances)
*Two-tailed
tests(Levene'stestusedto determine
Panel B: Students' Choices of Most FrequentlyUsed InstructionalMethods in PD
and TraditionalPrograms
Traditional
PD Program
Program
Instructional Method
(n = 62)
Case Analysis
InstructorLecture
Non-Case-Based Discussion
StructuredHomework Problems
77.4%
11.3%
9.7%
1.6%
4.5%
83.6%
3.6%
8.2%
100%
99.9%
Totals
(n= 110)
X2 (3) = 153.0, p <.001
Panel C: Students' Choices of Least FrequentlyUsed InstructionalMethods in PD
and TraditionalPrograms
PD Program
Traditional
(n = 62)
Program (n = 110)
InstructionalMethod
Case Analysis
InstructorLecture
Non-Case-Based Discussion
StructuredHomework Problems
1.6%
70.9%
27.4%
17.7%
53.2%
-?-%
19.1%
10.0%
Totals
99.9%
100%
X2(3) = 112.8,p< .001
Panel D: Students' Perceptions of Amountsof WorkRequired in PD and
TraditionalProgram Courses
PD Program
Traditional
(n = 62)
Program (n = 108)
InstructionalMethod
Far Too Little and Too LittleWork*
About the RightAmount of Work
Too Much Work
Far Too Much Work
Totals
4.8%
56.5%
35.5%
3.2%
2.8%
63.9%
30.5%
2.8%
100%
100%
X2 (3) = 3.2, p = .534
* Twocells("FarToo LittleWork"and "TooLittleWork")werecombineddue to smallcell sizes.
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
EDUCATING
FOR ACCOUNTING
EXPERTISE
49
We computed the composite variable representingprogramselection
characteristics(i.e., the Inverse Mills ratio) using fittedvalues estimated
fromthismodel.14
5. Results
5.1
PROGRAM
CHOICE
RESULTS
The probitmodel estimatingthe effectsof students'programchoices
has an overall predictionrate of 71.5% and a Maddala R2 = .21. These
measures of fitare similarto those of Berndt [1991] (Maddala R2 = .16;
predictionrate = 68%) and Pindyckand Rubinfeld [1991] (Maddala R2
= .185; prediction rate = 73.7%). The model indicates that students
were more likelyto enroll in the PD program if they: (X1 and X2) believed accountingwas interesting,(X3) believed thataccountingwas not
highlystructured,(X4) took high school accounting, (X5) had a higher
preaccountingUIUC GPA,(X6) were male, (X7) had higher GSE.15We
presentsubsequent ANCOVAresultsusing the InverseMills ratio derived
from this model to adjust for potential bias in estimated treatment
effectsdue to students'programchoices (Maddala [1983; 1991]).
5.2
INSTRUCTIONAL
PROCESS
RESULTS
Our firsttestof Hypothesis1, whichpredictsthatPD courses will contain more field projects and cases, group work,and presentationsthan
traditionalprogram courses, uses data fromPD and traditionalcourse
syllabi.Table 2, panel A compares the percentagesof points allocated to
field projects and cases versus structuredexaminations and quizzes in
the core PD and comparable traditionalprogram course syllabi. Core
PD courses weighted field projects and cases more heavily (t(19) = 6.0,
p < .001), while traditionalprogram courses weighted structuredexaminations and quizzes more heavily (t(19) = 5.7, p < .001). These data
indicate that, consistentwith PD programmaticgoals, the extrinsicincentivesofferedin the PD and traditionalprogramswere for differing
typesof assignments.16
14Using maximum likelihood probit estimation techniques, the fittedvalues in the
above model (Yi) are the probabilitiesthat studentschose Project Discovery,<D(X,); assuming that the associated standard normal densityfunctionis (p(X ), the resultingMills
W) forstudentschoosing Project Discovery,and [
forstudents
Li -(X~)
D (X3)i
choosing the traditionalprogram.The InverseMills ratiosthatwe use to testand adjust for
ratio isI[
L
1
)
(
LD(Xo) and LD(XJ3)1-
the effectsof students'programchoices are [ ((
respectively.
esecivly
15In addition, second-yearstudents' GSE scores explained less of the variance in their
students(X8).
programchoices than did the GSE scores of first-year
16To investigatepossible instructordifferencesin the twoprograms,we compared the 9
PD programinstructorswith9 of the 13 traditionalprograminstructors.Using two-tailed
we found no differencesbetween the PD and traditionalfacultyin age, number of
t-tests,
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
50
DAN N. STONE AND MARJORIE K. SHELLEY
As a second and a third test of Hypothesis 1, we asked graduates to
identifythe most and least frequentlyused instructionalmethod in the
PD (for PD graduates) and traditional(for traditionalprogram graduates) programs.Seventy-sevenpoint four percent (77.4%) of PD program graduates identified case analysis as the most frequentlyused
instructionalmethod, while 83.6% of traditional program graduates
identifiedinstructorlectures (see table 2, panel B). For the least frequently used instructionalmethods in PD and traditional program
courses, 53.2% of PD graduates identifiedstructuredhomeworkproblems as the least frequentlyused method, while 70.9% of traditional
programgraduatesidentifiedcase analysis(see table 2, panel C). X2tests
indicate statisticallysignificantdifferencesin cell means for both the
most and least frequentlyidentifiedinstructionalmethods.
Graduates' perceptions of the PD and traditionalprogramsmightbe
affectedby differingperceptions of the amounts of work required in
each program. However, results indicate no significantdifferencesin
graduates' perceptions of the amount of work required in the PD and
traditionalprograms (X2 (3) = 3.2, p = .534) (see table 2, panel D).17 In
both programs,more than half the participantsbelieved that accountancy departmentcourses required "about the rightamount of work,"
and about one-thirdbelieved the courses required "too much work."
As a finaltestof Hypothesis1, we analyzed responses to questions asking PD and traditionalprogram graduates the frequencywith which
theyworked in groups and made in-classpresentations.Panels A and B
of table 3 present these resultswithoutand with the Inverse Mills ratio
adjustmentfor participants' program choices. The responses indicate
that PD graduates more frequentlyworked in groups and made more
presentationsin accountancy classes; the statisticalsignificanceof the
resultsis unaffectedby our adjustmentforstudents'programchoices.
5.3 DECLARATIVE KNOWLEDGE
RESULTS18
5.3.1. CPAExaminationData. Panels A and B of table 4 present the
CPAexaminationdeclarativeknowledgeresultswithoutand withthe Inverse Mills ratio adjustmentforparticipants'programchoices. On average, PD graduates' CPAexaminationscores on the auditingsection were
years in a facultyposition, number of research grantsor awards received, or total number of undergraduate and Ph.D. courses taught previously(p > .17). On average, PD instructorshad higher academic rank, had received more teaching awards, had more
academic and total publications, and had served on more Ph.D. committees(p < .05). In
addition, tworesultsapproached conventional levels of statisticalsignificance:PD faculty
were more likelyto have doctorates (p < .07) and had taughtmore master'slevel courses
(p ' .09).
17There was no reliable differencebetween the PD and traditionalprogramsin survey
responses comparing the amount of workrequired in accountancycourses versuscourses
offeredby other departments(p > .15).
18We use the same set of explanatoryvariables,or a subset of them,forall subsequent
analyses. The set includes an instructionalapproach grouping variable, the Inverse Mills
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
EDUCATING
FOR ACCOUNTING
TABLE
51
EXPERTISE
3
Panel A: InstructionalProcess ANCOVAResults withoutProgram Choice Adjustment
p
F-Value
I frequentlyworked in groups in
the accounting classes I took at the
Universityof Illinois.
InstructionalMethod (PD or Traditional)
UIUC- GPA
ACT ComprehensiveScore
n= 138
Adj RX= 39%
I frequentlymade presentationsin
the accounting classes I took at the
Universityof Illinois.
InstructionalApproach (PD or Traditional)
UIUC- GPA
ACT ComprehensiveScore
n= 138
Adj. RX 61%
81.67
0.20
0.08
< .001
.656
.784
202.79
0.10
0.07
< .001
.759
.796
Panel B: InstructionalProcess ANCOVAResults withProgram Choice Adjustment
I frequentlyworked in groups in
the accounting classes I took at the
Universityof Illinois.
InstructionalMethod (PD or Traditional)
Inverse Mills Ratio
UIUC - GPA
ACT ComprehensiveScore
n = 136
Adj. R = 39%
I frequentlymade presentationsin
the accounting classes I took at the
Universityof Illinois.
InstructionalMethod (PD or Traditional)
Inverse Mills Ratio
UIUC-GPA
ACT ComprehensiveScore
n= 136
Adj. R2 = 62%
F-Value
p
41.71
0.36
0.21
0.05
<.001
.551
.644
.832
114.29
0.06
0.01
0.04
<.001
.815
.914
.850
higherthan those of traditionalprogramgraduates (PD graduates' average score = 74.2, traditionalprogramgraduates' average score = 70.5).19
Including the Inverse Mills ratio in the model reduces, but does not
eliminate, the effectof instructionalmethod on the auditing section
score. We calculated
ratio, UIUC GPA,ACT composite score, and a general self-efficacy
collinearitydiagnosticson these variables and found thatthe largestcondition index was
42. This value is large enough to cause concern about degraded regressionestimatesfor
variables involvedin the dependency. However, the variables involvedwere the constant,
UIUC GPA,and ACT composite score. The presence of the constantindicates thatthe two
other variables involved could each contain large invariantcomponents. Therefore,we
recomputed the collinearitydiagnostics aftersubtractingthe truncated means fromthe
UIUC GPA and ACT measures. These results indicated no serious collinearityproblems
(e.g., largestcondition index = 4).
19The average CPAexaminationpass ratesof both PD (43.8%) and traditionalprogram
candidates (Gleim
graduates (41.9%) exceed the national averages (15.5%) for first-time
[1996]).
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
52
DAN N. STONE AND MARJORIE K. SHELLEY
TABLE
4
Panel A: Declarative Knowledge CPAExaminationANCOVAResults withoutProgram
Choice Adjustment
CPAExamination Section
Audit
InstructionalMethod (PD or Traditional)
CPAReviewCourse Grade and Type
ACT ComprehensiveScore
UIUC- GPA
Financial Reporting
InstructionalMethod (PD or Traditional)
CPAReview Course Grade and Type
ACT ComprehensiveScore
UIUC-GPA
Business Law
InstructionalMethod (PD or Traditional)
CPAReviewCourse Grade and Type
ACT ComprehensiveScore
UIUC-GPA
Tax/Managerial/Governmental
InstructionalMethod (PD or Traditional)
CPAReviewCourse Grade and Type
ACT ComprehensiveScore
UIUC- GPA
n = 162
Adj. R2 = 56%
n = 162
Adj. R2 = 48%
n = 162
Adj. R2 = 40%
n= 162
Adj. R2 = 55%
F-Value
p
6.18
12.73
14.66
43.44
.014
<.001
<.001
<.001
1.71
5.53
7.29
52.06
.193
.005
.008
<.001
.006
4.30
7.67
32.22
.939
.015
.006
<.001
3.46
8.54
5.96
63.39
.065
<.001
.016
<.001
Panel B: Declarative Knowledge CPAExaminationANCOVAResults withProgram
Choice Adjustment
CPAExamination Section
Audit
InstructionalMethod (PD or Traditional)
CPAReviewCourse Grade and Type
ACT ComprehensiveScore
UIUC- GPA
Inverse Mills Ratio
Financial Reporting
InstructionalMethod (PD or Traditional)
CPAReview Course Grade and Type
ACT ComprehensiveScore
UIUC- GPA
Inverse Mills Ratio
F-Value
n= 144
Adj. R2 = 55%
n = 144
Adj. R2 = 48%
p
4.94
14.22
6.08
28.76
.039
.028
<.001
.015
<.001
.844
.717
4.40
6.24
48.86
5.10
.399
.014
.014
<.001
.026
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
EDUCATING
TABLE
FOR ACCOUNTING
53
EXPERTISE
4-continued
Business Law
InstructionalMethod (PD or Traditional)
CPAReview Course Grade and Type
ACT ComprehensiveScore
UIUC- GPA
Inverse Mills Ratio
Tax/Managerial/Governmental
InstructionalMethod (PD or Traditional)
CPAReview Course Grade and Type
ACT ComprehensiveScore
UIUC-GPA
Inverse Mills Ratio
n = 144
Adj. R2 = 37%
n= 144
Adj. R2 = 54%
.91
4.37
5.69
21.67
1.63
.343
.014
.018
<.001
.203
.16
4.88
4.06
60.69
2.31
.694
.009
.046
<.001
.131
scores. There were no reliable differencesdue to instructionalmethod
on the other sectionsof the CPAexamination (p > .343).
5.4
INTELLECTUAL
SKILL RESULTS
Hypothesis3 predictsthatintellectualskillsmeasureswillbe higherfor
PD than for traditionalprogram graduates. Table 5, panel A compares
PD and traditionalprogramgraduates' intellectualskillmeasures before
the program choice adjustment;all four analyses show a significantinstructional method effect.Panel B shows the same comparisons after
the adjustment.The second set of resultsindicates that,for two of the
informationrefourmeasures (i.e., cognitivecomplexityand identifying
sources), instructionalapproach does not explain differencesin PD and
traditionalgraduates beyond those accounted for by program choice
characteristics.However,instructionalapproach explains significantadskillsbeyond thatexplained by
ditional variance in problem-structuring
the Inverse Mills ratio alone. In addition, the additional variance explained by the effectof instructionalapproach on writtencommunication skillsapproaches conventionalsignificancelevels.
5.5
ATTITUDE
RESULTS
Hypothesis4 predictsthatPD graduateswill have higher tolerance for
ambiguitymeasures (H4a) and will rate accountancy department instructionmore highly (H4b) than traditionalprogram graduates. The
resultsin table 6, panel A indicate that differencesbetween groups on
the tolerance for ambiguitymeasure resultfromstudentcharacteristics
and programchoice ratherthan instructionalmethod.20Consistentwith
the lack of differencesin declarativeknowledge,there are no betweengroup differencesin attitudesrelated to increases in accounting knowledge. Other attitude measures indicate instructionaltreatmenteffects
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
54
DAN N. STONE AND MARJORIE K. SHELLEY
TABLE
5
Panel A: Intellectual Skill ANCOVAResults withoutProgram Choice Adjustment
Intellectual Skill Measure
CognitiveComplexity
InstructionalMethod (PD or Traditional)
UIUC- GPA
ACT Composite Score
IdentifyingInformationResources
InstructionalMethod (PD or Traditional)
UIUC-GPA
ACT Composite Score
Problem Structuring
InstructionalMethod (PD or Traditional)
UIUC-GPA
ACT ComprehensiveScore
WrittenCommunication Skills
InstructionalMethod (PD or Traditional)
UIUC- GPA
ACT ComprehensiveScore
F-Value
P
n= 132
Adj. R2 = 8%
n= 122
Adj. R2= 3%
3.19
.000
6.97
.038*
.987
.009
5.11
.55
.29
.013*
.460
.592
17.79
4.16
.22
<.001*
.044
.637
10.98
7.31
.41
<.001*
.008
.562
n = 122
Adj. R2 = 13%
n= 122
Adj. R2 = 10%
Panel B: Intellectual SkillANCOVAResults withProgram Choice Adjustment
Intellectual Skill Measure
CognitiveComplexity
InstructionalMethod (PD or Traditional)
UIUC- GPA
ACT Composite Score
Inverse Mills Ratio
IdentifyingInformationResources
InstructionalMethod (PD or Traditional)
UIUC- GPA
ACT Composite Score
Inverse Mills Ratio
Problem Structuring
InstructionalMethod (PD or Traditional)
UIUC- GPA
ACT Composite Score
Inverse Mills Ratio
WrittenCommunication Skills
InstructionalMethod (PD or Traditional)
UIUC-GPA
ACT Composite Score
Inverse Mills Ratio
p
F-Value
n= 129
Adj. R2 = 7%
n = 115
Adj. R2 = 4%
1.49
.000
7.36
.007
.112*
.997
.008
.934
.443
.647
.383
2.13
.254*
.423
.537
.148
6.18
3.61
.095
.226
.007*
.060
.758
.636
2.43
7.49
.435
.961
.061*
.007
.511
.329
n = 115
Adj. R2 = 11%
n = 115
Adj. R2 = 9%
*One-tailed
tests.
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
EDUCATING
FOR ACCOUNTING
EXPERTISE
55
and are relativelyinsensitiveto our adjustmentforprogramchoice characteristics.For example, PD graduates more frequentlyagreed that "the
accountancyfacultyare excellent teachers"and that theywere "pleased
withthe accounting education" theyreceived at the Universityof Illinois
than did traditionalprogramgraduates.
6. Discussion
Our resultssuggestinstructionalprocess differencesbetween the PD
and traditionalprogramsand learning outcome and attitudedifferences
between PD and traditionalprogram graduates. Specifically,analysisof
syllabi and surveys of graduates indicate that PD instructionmade
greater use of expertise-buildinginstructionalprocesses such as field
projects,case analysis,group work,and presentations(Hi). In contrast,
the traditional program made greater use of instructorlectures and
weightedexaminationsand surveysmore heavilythan the PD program.
The data suggestfewdifferencesbetween PD and traditionalprogram
graduates (H2) in theirperformanceon the CPAexamination; however,
PD graduates achieved significantlyhigher auditing scores, even after
adjusting for pre-program selection characteristics.PD students also
show evidence of strongerproblem-structuring
skills and better attitudes afteradjusting for program choice characteristics.However, observed differencesin cognitive complexity,writing,and informational
retrievalappear to resultmore fromprogramchoice characteristicsthan
instructionalmethod.21
A trade-offin designing the research was whether to use contextindependent or accounting-relevant(i.e., context-dependent)measures
of the dependent variables.Cognitiveresearch suggeststhatknowledge,
skill,and expertiseare contextdependent; because of this,we chose to
use accounting-relevantmeasures of all except one (tolerance for ambiguity) of our post-programdependent variables. But a disadvantage
20We also collected measures of tolerance forambiguityat the beginningof thejunior
year and at the end of the senior year for these graduates. Analysisof the two measures
indicates that PD graduates' tolerance for ambiguitymeasures did not change, whereas
traditionalprogramgraduates' tolerance forambiguitymeasures declined (p = .015).
21 One potential explanation for the observed improvementsin performanceamong
PD graduates is the existence of a "Hawthorneeffect,"arisingbecause PD graduateswere
aware that theywere participatingin an experimentalprogram (Adair [1984] and Simon
and Burstein [1985]). Baron [1987] argues that informingcontrol groups that theyalso
are receivingspecial treatmentminimizesHawthorne effects.We implemented this suggestion throughperiodic remindersto studentsin the traditionalprogram (via newsletters,outside speakers,remarksby the departmenthead, etc.) thattheywere enrolled in an
excellent, top-rankedaccountancyprogram.Althoughwe attemptedto control for possible Hawthorne effects,considerable evidence suggests that Hawthorne effectsare rare
and, when present,of small magnitude (Cook and Campbell [1979]). In addition, more
recentresultshave led severalauthorsto argue that"Hawthorneeffects"are sufficiently
illdefined and illusive as to be an unusable research construct (Adair, Sharp, and Huynh
[1989] and Diaper [1990]).
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
56
DAN N. STONE AND MARJORIE K. SHELLEY
TABLE
6
Panel A: AttitudeANCOVAResults withoutProgram Choice Adjustment
AttitudesTowardInstruction
F-Value
Tolerance forambiguity.
n = 142
Adj. R2 = 5%
InstructionalMethod (PD or Traditional)
2.27
UIUC-GPA
.949
General Self-Efficacy
7.24
ACT Composite Score
2.29
Accountancycourses improvedmy
communicationskills.
InstructionalMethod (PD or Traditional)
UIUC- GPA
General Self-Efficacy
ACT Composite Score
n= 138
Adj. R2 = 39%
Accountancycourses improvedmy
interpersonalskills.
InstructionalMethod (PD or Traditional)
UIUC- GPA
General Self-Efficacy
ACT Composite Score
n = 138
Adj. R2 = 29%
Accountancyimprovedmycreativityand
problem-solvingabilities.
InstructionalMethod (PD or Traditional)
UIUC-GPA
General Self-Efficacy
ACT Composite Score
n= 138
Adj. R2 = 27%
The accountancyfacultyare excellent
teachers.
InstructionalMethod (PD or Traditional)
UIUC- GPA
General Self-Efficacy
ACT Composite Score
n = 138
Adj. R2 = 10%
I am pleased withthe accounting education
I received at the Universityof Illinois.
InstructionalMethod (PD or Traditional)
UIUC- GPA
General Self-Efficacy
ACT Composite Score
n = 138
Adj. R2 = 11%
Accountancycourses improvedmy
accounting knowledge.
InstructionalMethod (PD or Traditional)
UIUC- GPA
General Self-Efficacy
ACT Composite Score
n = 138
Adj. R2 = 2%
p
.067*
.332
.008
.133
88.45
.797
4.68
.053
<.001*
.374
.032
.818
53.25
.221
5.26
.493
<.001*
.639
.023
.484
45.94
3.31
5.05
1.77
<.001*
.071
.026
.186
14.37
.345
1.42
.40
<.001*
.558
.236
.528
4.62
2.41
3.41
2.16
.017*
.123
.067
.144
.40
.37
1.81
4.88
.736*
.547
.181
.029
Panel B: AttitudeANCOVAResults withProgram Choice Adjustment
AttitudesTowardInstruction
F-Value
Tolerance forambiguity.
InstructionalMethod (PD or Traditional)
UIUC- GPA
n = 140
Adj. R2 = 5%
.21
.47
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
p
.326*
.492
EDUCATING
TABLE
FOR ACCOUNTING
57
EXPERTISE
6-continued
General Self-Efficacy
ACT Composite Score
Inverse Mills Ratio
Accountancycourses improvedmy
communicationskills.
InstructionalMethod (PD or Traditional)
UIUC- GPA
General Self-Efficacy
ACT Composite Score
Inverse Mills Ratio
n= 136
Adj. R2 = 40%
Accountancycourses improvedmy
interpersonalskills.
InstructionalMethod (PD or Traditional)
UIUC- GPA
General Self-Efficacy
ACT ComprehensiveScore
InverseMills Ratio
n= 136
Adj. R2 = 32%
Accountancyimprovedmycreativityand
problem-solvingabilities.
InstructionalMethod (PD or Traditional)
UIUC- GPA
General Self-Efficacy
ACT Composite Score
Inverse Mills Ratio
n= 136
Adj. R2 = 28%
The accountancyfacultyare excellent
teachers.
InstructionalMethod (PD or Traditional)
UIUC- GPA
General Self-Efficacy
ACT Composite Score
Inverse Mills Ratio
n = 136
Adj. R2 = 10%
I am pleased withthe accounting education
I received at the Universityof Illinois.
InstructionalMethod (PD or Traditional)
UIUC- GPA
General Self-Efficacy
ACT Composite Score
InverseMills Ratio
n = 136
Adj. R2 = 11%
Accountancycourses improvedmy
accounting knowledge.
InstructionalMethod (PD or Traditional)
UIUC- GPA
General Self-Efficacy
ACT Composite Score
Inverse Mills Ratio
*One-tailed
tests.
n= 136
Adj. R2 = 4%
7.13
2.31
1.29
.009
.131
.259
43.96
.67
4.02
.090
.90
<.001*
.414
.047
.765
.345
21.41
.05
4.71
.43
3.00
<.001*
.816
.032
.512
.086
20.45
2.89
4.65
1.72
1.01
<.001*
.092
.033
.193
.318
7.82
.25
1.09
.30
.059
.003*
.616
.299
.585
.808
.64
2.92
3.24
2.22
1.87
.213*
.090
.074
.139
.174
3.30
.10
2.08
5.62
4.28
.964*
.753
.151
.019
.041
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
58
DAN N. STONE
AND
MARJORIE
K. SHELLEY
of our approach is that collecting pretest values on the dependent
measures is problematic.Specifically,administeringpretestmeasures on
our accounting-relevant
variableswould have led us to ask first-semester
accounting studentsto answer questions forwhich most would have no
basis for responding.22Because we do not have pretestscores on our
declarative knowledge and intellectual skills measures, we cannot say
whether these measures changed during the accounting program or
differedinitiallybetween the PD and traditionalgroups.
Assuming that the observed effectsresult from instructionaldifferences, will they generalize to other settings?One potential threat to
the generalizabilityof the resultsis the consistentlyhigh qualityof students admitted to the undergraduate accountancy program at the
UIUC; some argue that PD is essentiallyan honors program,the results
of which will not generalize to less homogeneously adept students.A
related threat to the generalizabilityof the results is our inabilityto
identifythe specificcausal mechanisms that are responsible for the efdifferenceswe note
fectswe identify.Specifically,the instruction-related
may resultfromone or more of the followingfactors,or fromspecific
combinationsor interactionsof these factors:(1) PD instructors,(2) the
"organizationalculture" created by the PD initiative,(3) characteristics
of studentswho enrolled in the PD program,(4) course content of the
PD curriculum,(5) instructionalmethodsof the PD program,(6) incentivesofferedto instructors,(7) incentivesofferedto students,or (8) perceptions of other program stakeholders (e.g., program administrators,
recruiters,accounting professionals).While we believe that curriculum
content and instructionalmethods influenced the reported results,our
research design does not allow us to rule out competingexplanations.
REFERENCES
CHANGE COMMISSION
(AECC). "Objectives of Education for
EDUCATION
Education5, no. 2
Accountants:Position StatementNumber One." Issues in Accounting
(Fall 1990): 307-12.
Berkeley:Universityof California
AnalysisofQuasi-Experiments.
ACHEN, C. H. The Statistical
Press, 1986.
ADAIR, J. G. "The Hawthorne Effect:A Reconsideration of the Methodological Artifact."
69 (1984): 334- 45.
JournalofAppliedPsychology
"Placebo, Hawthorne,and Other ArtifactConADAIR, J. G.; D. SHARPE; AND C. HUYNH.
Education57, no. 4
trols: Researchers' Opinions and Practices."JournalofExperimental
(1989): 341-55.
AKIN, M. C., AND ASSOCIATES.A GuideforEvaluationDecisionMakers.London: Sage, 1985.
edited by C. Murchison.
ALLPORT,
G. W. "Attitudes."In Handbookof Social Psychology,
Worcester,Mass.: Clark UniversityPress, 1935.
and Thought.Hillsdale, N.J.: Erlbaum, 1976.
ANDERSON, J. R. Language,Memory,
ACCOUNTING
22Subsequently,we attemptedunsuccessfullyto collect such measures. Attemptingto
collect pretestmeasures of accounting-relevantknowledgeand skillproduced manyfrivolous and unresponsiveanswers (e.g., "I like baseball") and a large majorityof responses
demonstratingno relevantknowledge.
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
EDUCATING
FOR ACCOUNTING
EXPERTISE
59
. Cognitive
Psychology
and Its Implications.
2d ed. New York:Freeman, 1985.
BAKER, R.; B. CUNNINGHAM; P. KIMMELL; AND C. VENABLE. A CatalogofResource
Materialsfor
TeachingAccounting
StudentsCriticalThinking,
Accounting
PedagogicalResourceSeries,vol. 1,
edited by K. Pincus. St. Louis: Federation of Schools of Accountancy,1995.
BARON, J.B. "Questions and Answersabout the Nature and Teaching of ThinkingSkills."
In Teachingof ThinkingSkills:Theoryand Practice,edited by J.B. Baron and R. J.Sternberg, pp. 221-48. New York:Freeman, 1987.
BARTUNEK, J.M.; J.R. GORDON; AND R. P. WEATHERSBY. "Developing Complicated Understandingin Administrators."
AcademyofManagement
Review8, no. 2 (1983): 273-84.
and Society
BIDARD, J."Expertisein Auditing:Mythor Reality."Accounting,
Organizations,
14,
nos. 1/2 (1989): 113-31.
"Expertise and Its Relation to Audit Decision Quality." Contemporary
Accounting
Research8, no. 1 (1991): 198-222.
BERNDT, E. R. ThePracticeofEconometrics.
Reading, Mass.: Addison-Wesley,1991.
BIG EIGHT ACCOUNTING FIRMS (Arthur Andersen and Co.; Arthur Young; Coopers and
Lybrand;Deloitte, Haskins, and Sells; Ernstand Whinney;Peat MarwickMain and Co.;
Price Waterhouse; and Touche Ross). Perspectives
on Education:Capabilities
for Successin
theAccounting
New York,1989.
Profession.
BLUMBERG, M., AND C. D. PRINGLE. "How Control Groups Can Cause Loss of Control in
ActionResearch: The Case of Rushton Coal Mine."JournalofAppliedBehavioralScience19
(1983): 409-25.
in theClassroom,
ASHE-ERIC
Excitement
BONWELL, C., AND J. EiSON. ActiveLearning:Creating
HigherEducationReport1. Washington,D.C.: ERIC Clearinghouse on Higher Education,
George WashingtonUniversity,1991.
BONNER, S. E., AND N. P. PENNINGTON. "Cognitive Processes and Knowledge as Determi10 (1991): 1-50.
nants of AuditorExpertise."JournalofAccounting
Literature
BONNER, S. E., AND P. L. WALKER. "The Effects of Instruction and Experience on the
Review69, no. 1 (1994): 157-78.
Acquisitionof AuditingKnowledge." TheAccounting
BREUER, K., AND R. D. TENNYSON. "PsychologicalFoundations for InstructionalDesign
Theory: CognitiveComplexityTheory."Journalof Structured
Learning12, no. 3 (1995):
165-73.
BRUNSWIK, E. Perception
and theRepresentative
Design of Psychological
Experiments.
Berkeley:
Universityof CaliforniaPress, 1956.
CHEN, K. H., AND S. K. OLSON. "Measuring Cognitive Complexity in the Accounting Do1 (1989): 160-81.
main." BehavioralResearchin Accounting
CHI, M.; R. GLASER; AND M. FARR. TheNatureofExpertise.
Hillsdale, N.J.: Erlbaum, 1988.
COOK, T., AND D. CAMPBELL. Quasi-Experimentation:
Design and AnalysisIssuesforField Settings.Boston: Houghton Mifflin,1979.
DEMONG, R. F.; J. H. LINDGREN, JR.; AND S. E. PERRY. "Designing an Assessment Program
forAccounting."Issuesin Accounting
Education9, no. 1 (1994): 11-27.
DEPPE, L. A.; E. 0. SONDEREGGER; J. D. STICE; D. C. CLARK; AND G. F STREULING.
Educa"EmergingCompetencies for the Practice of Accountancy."JournalofAccounting
tion9 (1991): 157-90.
DIAPER, G. "The Hawthorne Effect:A Fresh Examination." Education Studies16, no. 3
(1990): 261-67.
to Increase VolunteerEDEN, D., AND J. KINNAR. "Modeling Galatea: Boosting Self-Efficacy
76 (1991): 770-80.
ing."JournalofAppliedPsychology
EDUCATIONAL TESTING SERVICE. Guidelines
Tests.
for Developingand ScoringFree-Response
Princeton,N.J.: Educational TestingService, 1987.
ENNIS, R. H. "A Taxonomy of Critical Thinking Dispositions and Abilities."In Teaching
ThinkingSkills:Theoryand Practice,edited by J. B. Baron and R. J. Sternberg,pp. 1-26.
New York:Freeman, 1987.
of Learning fromExperience." In The ThinkingOrganizaFELDMAN, J. "On the Difficulty
tion,edited by H. P. Sims, Jr.,and D. A. Gioia, pp. 263-92. San Francisco: Jossey-Bass,
1986.
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
60
DAN N. STONE
AND
MARJORIE
K. SHELLEY
FOWLES,M. "Holistic Scoring witha Published Scoring Guide." Technical memorandum,
Educational TestingService, 1990.
FRANCIs,M. C.; T. C. MULDAR;AND J. S. STARK.IntentionalLearning:A Process
forLearning
toLearnin theAccounting
Curriculum.
AccountingEducation Series, no. 12. Sarasota, Fla.:
AmericanAccountingAssn., 1995.
FRENSCH,P A., AND R. J. STERNBERG."Expertise and IntelligentThinking: When Is It
edited by R. J.
Worse to Know Better?"In Advancesin thePsychology
ofHuman Intelligence,
Sternberg,pp. 157-88. Hillsdale, N.J.: Erlbaum, 1989.
GAGNE, R. M. "LearningOutcomes and Their Effects:Useful Categories of Human Performance." AmericanPsychologist
39, no. 4 (1984): 377-85.
GAINEN, J. "Developing Intellectual Skills in AccountingEducation." In CriticalThinking
Interactive
Learning,and Technology:
ReachingforExcellencein BusinessEducation,edited by
T. J. Frecka. New York:ArthurAnderson & Co., 1992.
New York:Harper Collins,
GARDNER, H. FramesofMind: The Theory
ofMultipleIntelligences.
1983.
Mind: How ChildrenThinkand How SchoolsShouldTeach.New York:
_ . The Unschooled
Basic Books, 1991.
. EducatingtheUnschooled
Mind. Science and Public PolicySeminar at the Federation
of Behavioral, Psychological,and Cognitive Sciences, Occasional Paper. Washington,
D.C.: Federation of Behavioral, Psychological,and CognitiveSciences, 1993.
GLASER, R. TheNatureofExpertise.
Science and Public PolicySeminar at the Federation of
Behavioral, Psychological,and CognitiveSciences, Occasional Paper no. 107. Washington, D.C.: Federation of Behavioral,Psychological,and CognitiveSciences, 1989.
GLEIM, I. N. CPA Review:A System
forSuccess.Gainesville,Fla.: Gleim Publications,1996.
GRE FACULTY ADVISORYTEAM. "GRE WritingScoring Guide." Technical manual, Educational TestingService, 1991.
GUPTA,P. P., AND T. J. FOGARTY."GovernmentalAuditorsand Their Tolerance forAmbiguity:An Examinationof the Effectsof a PsychologicalVariable." Government
Accountants
Journal(Fall 1993): 25-35.
HAMMOND,K. R. "ProbabilisticFunctionalism:Egon Brunswik'sIntegrationof the History,
Theory, and Method of Psychology."In The Psychology
ofEgon Brunswik,
by K. R. Hammond, pp. 1-66. New York:Holt, Rinehart& Winston,1966.
KMENTA, J. Elements
ofEconometrics.
New York:Macmillan, 1986.
KOCH, J. L., AND S. R. RHODES.
"ProblemswithReactive Instrumentsin Field Research."
JournalofAppliedBehavioralScience15 (1979): 485-506.
and Human Information
LIBBY, R. Accounting
Processing:Theoryand Applications.
Englewood
Cliffs,N.J.: Prentice-Hall,1981.
and Deci_ . "The Role of Knowledge and Memoryin Audit Judgment."In Judgment
sion MakingResearchin Accounting
and Auditing,edited by R. H. Ashton and A. H. Ashton, pp. 176-206. New York:Cambridge UniversityPress, 1995.
LIBBY, R., AND H. T. TAN. "Modeling the Determinantsof Audit Expertise."Accounting,
and Society19, no. 8 (1994): 701-16.
Organizations,
and QualitativeVariables
in Econometrics.
MADDALA, G. S. Limited-Dependent
Cambridge: Cambridge UniversityPress, 1983.
. "A Perspectiveon the Use of Limited-Dependentand QualitativeVariables Models
_
in AccountingResearch." TheAccounting
Review66, no. 4 (1991): 788-807.
A. P., JR. "Revised Scale for AmbiguityTolerance: Reliabilityand Validity."
MAcDONALD,
Psychological
Reports
26 (1970): 791-98.
MARCHANT, G. "Discussion of Determinantsof Auditor Expertise."JournalofAccounting
Research(Supplement 1990): 21-28.
PINCUS, K. V. "AuditorIndividual Differencesand Fairness of PresentationJudgments."
Auditing:A JournalofPracticeand Theory
9, no. 3 (1990): 150-66.
3 (1991): 39-65.
in Accounting
_ . "AuditJudgmentConfidence."BehavioralResearch
PINDYCK, R., AND D. RUBINFELD.
Econometric
Models and EconomicForecasts.3d ed. New
York:McGraw-Hill,1991.
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions
EDUCATING
FOR ACCOUNTING
EXPERTISE
61
W. GettingSmartQuicker:TrainingMoreSkillsin Less Time.Science and Public
Policy Seminar at the Federation of Behavioral Psychological and Cognitive Sciences,
Occasional Paper. Washington,D.C.: Federation of Behavioral, Psychological,and CognitiveSciences, 1989.
SHEHATA, M. "Self-SelectionBias and Economic Consequences of AccountingRegulation:
An Application of Two-StageSwitchingRegressionto SFAS No. 2." TheAccounting
Review
66, no. 4 (1991): 768-87.
SHERER, M.; J. E. MADDUX;
B. BERCANDANTE; S. PRENTICE-DUNN;
B. JACOBS; AND R. W.
ROGERS. "The Self-Efficacy
Scale: Constructionand Validation."Psychological
Reports51
(1982): 663-71.
SIMON, J. L., AND P. BURSTEIN.
Basic ResearchMethodsin Social Science.3d ed. New York:
McGraw-Hill,1985.
SOLOMON,
I. "Project Discovery."In Accounting
EducationChange:BestPractices,edited by
R. Flaherty,Accounting Education Series, vol. 14. Sarasota, Fla.: American Accounting
Assn., 1997.
New York:AcaSTREUFERT, S., AND R. W. SWEZEY. Complexity,
Managers,and Organizations.
demic Press, 1986.
D. L., AND A. J. SHINKFIELD. Systematic
Evaluation.Boston, Mass.: KluwerSTUFFLEBEAM,
Nijhoff,1985.
SUEDFELD,
P.; P. E. TETLOCK; AND S. STREUFERT. "Conceptual/Integrative
Complexity."In
Motivationand Personality:
Handbookof ThematicContentAnalysis,edited by C. P. Smith,
pp. 393-418. Cambridge: Cambridge UniversityPress, 1992.
"The Measurementof Generalized Self-Efficacy:
TIPTON, R. M., AND E. L. WORTHINGTON.
A Studyof ConstructValidity."JournalofPersonality
Assessment
48, no. 5 (1984): 545- 49.
SCHNEIDER,
UNIVERSITY
OF
ILLINOIS
AT
URBANA-CHAMPAIGN,
DEPARTMENT
OF
ACCOUNTANCY.
"ProjectDiscovery:A PrototypeforEducation in Accountancy."Programbrochure,Universityof Illinois, 1995.
WEST, C. K.; J.A. FARMER; AND P. M. WOLFF Instructional Design: Implicationsfrom Cognitive
Science.Boston: Allynand Bacon, 1991.
WINOGRAD,T. "Frame Representationsand the Declarative-ProceduralControversy."In
Representation
and Understanding:
Studiesin CognitiveScience,edited by D. Bobrow and
J. A. Collins, pp. 185-210. New York:Academic Press, 1975.
YEo, G. G. H., AND D. A. ZIEBART."The Effectof Self-SelectionBias on the Testing of a
Finance
Stock Price Reaction to Management's EarningsForecasts."ReviewofQuantitative
5 (1995a): 5-55.
and Accounting
. "An Empirical Test of the Signaling Effectof Management's Earnings Forecasts:
A Decomposition of the Earnings Surprise and Forecast SurpriseEffects."Journal
ofAccounting,
Auditing,and Finance10, no. 4 (1995b): 787-802.
This content downloaded from 152.20.158.206 on Wed, 30 Sep 2015 17:12:24 UTC
All use subject to JSTOR Terms and Conditions