the relationship of academic self-efficacy to class participation and exam performance

17
Soc Psychol Educ (2012) 15:233–249 DOI 10.1007/s11218-011-9175-x The relationship of academic self-efficacy to class participation and exam performance Charles E. Galyon · Carolyn A. Blondin · Jared S. Yaw · Meagan L. Nalls · Robert L. Williams Received: 19 December 2010 / Accepted: 4 November 2011 / Published online: 1 December 2011 © Springer Science+Business Media B.V. 2011 Abstract This study examined the relationship of academic self-efficacy to engagement in class discussion and performance on major course exams among stu- dents ( N = 165) in an undergraduate human development course. Cluster analysis was used to identify three levels of academic self-efficacy: high (n = 34), medium (n = 91), and low (n = 40). Results indicated that high, medium, and low aca- demic self-efficacy all significantly predicted levels of student participation and exam performance, but the directionality of group placement on the academic measures was different for students at the high self-efficacy level versus those at the low and mid self-efficacy levels. Cluster analysis was also used to divide students into high, medium, and low grade-point average (GPA). These groups did not differ significantly on either self-efficacy or class participation but did differ on exam performance. Within GPA levels, self-efficacy was most strongly related to class participation and exam performance at the highest level of GPA and least related at the lowest level of GPA. Keywords Self-efficacy · College · Exam performance · Participation · GPA 1 Introduction Self-efficacy is the term used to describe a person’s belief that he/she has the ability to perform a particular activity or behavior. Bandura (1977) initially proposed self-effi- cacy as an explanation for behavior change during psychotherapy. Self-efficacy has been found to be predictive of sports skills, health practices, socialization, and aca- demic performance (Owen and Froman 1988). In examining the predictive potential of academic self-efficacy, we first offer distinctions between self-efficacy and kindred C. E. Galyon (B ) · C. A. Blondin · J. S. Yaw · M. L. Nalls · R. L. Williams Department of Educational Psychology and Counseling, The University of Tennessee, Knoxville, TN 37996, USA e-mail: [email protected] 123

Upload: robert-l-williams

Post on 25-Aug-2016

216 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: The relationship of academic self-efficacy to class participation and exam performance

Soc Psychol Educ (2012) 15:233–249DOI 10.1007/s11218-011-9175-x

The relationship of academic self-efficacy to classparticipation and exam performance

Charles E. Galyon · Carolyn A. Blondin ·Jared S. Yaw · Meagan L. Nalls ·Robert L. Williams

Received: 19 December 2010 / Accepted: 4 November 2011 / Published online: 1 December 2011© Springer Science+Business Media B.V. 2011

Abstract This study examined the relationship of academic self-efficacy toengagement in class discussion and performance on major course exams among stu-dents (N = 165) in an undergraduate human development course. Cluster analysiswas used to identify three levels of academic self-efficacy: high (n = 34), medium(n = 91), and low (n = 40). Results indicated that high, medium, and low aca-demic self-efficacy all significantly predicted levels of student participation and examperformance, but the directionality of group placement on the academic measureswas different for students at the high self-efficacy level versus those at the low andmid self-efficacy levels. Cluster analysis was also used to divide students into high,medium, and low grade-point average (GPA). These groups did not differ significantlyon either self-efficacy or class participation but did differ on exam performance. WithinGPA levels, self-efficacy was most strongly related to class participation and examperformance at the highest level of GPA and least related at the lowest level of GPA.

Keywords Self-efficacy · College · Exam performance · Participation · GPA

1 Introduction

Self-efficacy is the term used to describe a person’s belief that he/she has the ability toperform a particular activity or behavior. Bandura (1977) initially proposed self-effi-cacy as an explanation for behavior change during psychotherapy. Self-efficacy hasbeen found to be predictive of sports skills, health practices, socialization, and aca-demic performance (Owen and Froman 1988). In examining the predictive potentialof academic self-efficacy, we first offer distinctions between self-efficacy and kindred

C. E. Galyon (B) · C. A. Blondin · J. S. Yaw · M. L. Nalls · R. L. WilliamsDepartment of Educational Psychology and Counseling,The University of Tennessee, Knoxville, TN 37996, USAe-mail: [email protected]

123

Page 2: The relationship of academic self-efficacy to class participation and exam performance

234 C. E. Galyon et al.

control variables: perceived skill versus actual skill, current skill versus future skill,self-efficacy versus self-concept, and process versus product criterion variables. Lastly,we examine the role of GPA as a possible moderator variable between self-efficacyand criterion measures.

1.1 Distinctions between self-efficacy and kindred variables

1.1.1 Perceived skill versus objective skill

An important distinction regarding self-efficacy is between the belief that one hasa particular skill versus objective confirmation of that skill. For that reason, someresearchers have claimed that self-efficacy predicts performance of a particular behav-ior only when one has the skill to perform that behavior. Chowdhury and Shahabuddin(2007) assert that self-efficacy alone will not insure success if skill is lacking. Indeed,Bandura’s (1977) original conceptualization of self-efficacy offered this interpretation.His conception was that self-efficacy may predict attempts to perform the designatedbehavior but not necessarily the skill level achieved. Therefore, specific achievementoutcomes are best conceptualized as a function of three characteristics: efficacy, skill,and will (McCombs and Marzano 1990).

1.1.2 Current skill versus future skill

Another distinction needing greater clarification in the self-efficacy literature iswhether self-efficacy primarily relates to the application of skills already in one’srepertoire, the future attainment of skills, or both. Bandura defined self-efficacy asthe “conviction that one can successfully execute the behavior …” (p. 193). However,it is unclear whether Bandura meant that this conviction targets a current skill or thepotential to acquire that skill. Perhaps self-efficacy subsumes both possibilities.

Nonetheless, the distinction between the perceived ability to currently perform aparticular skill versus the perceived ability to acquire that skill could make an impor-tant difference in judging one’s suitability for a role requiring that skill. For example,an employer considering someone for a job requiring particular skills would probablybe more inclined to hire a person who claims to already have the required skills thanone who claims s/he could acquire the skills. In other words, claiming to have specificskills may lead to a different outcome than claiming the potential to acquire the skills.

On the other hand, believing that one already has a particular skill versus believ-ing that he or she could acquire that skill may affect the amount of effort investedin acquiring or refining the skill. Vancouver and Kendall (2006) reported that highself-efficacy regarding one’s current ability to perform a particular task may resultin minimal effort to prepare for a task requiring that skill, which could subsequentlylead to poor performance on the task; whereas lower self-efficacy may result in one’sworking hard to prepare for the task and ultimately performing well on the task. On theother hand, if self-efficacy represents one’s perceived ability to acquire a skill not cur-rently in the person’s repertoire, high self-efficacy is likely to produce more extendedeffort to acquire the skill than low self-efficacy. Plus, the increased perseverance in

123

Page 3: The relationship of academic self-efficacy to class participation and exam performance

Academic self-efficacy 235

attempting to acquire the skill is likely to lead to better performance than equivocaland abbreviated effort to acquire the skill, as typically might be expected with lowself-efficacy.

But once a skill is acquired, high self-efficacy regarding the performance of thatskill may not produce better performance than low self-efficacy. In earlier research,Vancouver et al. (2001, 2002) claimed that self-efficacy was positively related to pastperformance (students believe they can perform the skill because they have previ-ously demonstrated the skill) but negatively related to subsequent execution of theskill (being confident they can perform the skill may result in minimal effort to main-tain that skill).

1.1.3 Self-efficacy versus self-concept

Items on a self-efficacy scale should measure belief in one’s ability to perform spe-cific skills, such as “computing a standard deviation,” rather than broad domain skills,such as “doing statistics” (Choi 2005; Finney and Schraw 2003). Researchers havefound that academic self-efficacy beliefs are organized hierarchically, such that overtime students first develop beliefs regarding their specific math skills (e.g., substi-tuting numbers for letters, solving for unknowns), then beliefs regarding their skillsin mathematical domains (e.g., algebra, calculus), and finally beliefs regarding theircapabilities in larger academic domains (e.g., mathematics, English) (Lent et al. 1997).A concept anchored in the belief that one can perform one or more specific skills canlegitimately be labeled self-efficacy. However, when the notion is broader and moreevaluative in nature (e.g., believing you are a good math student without referenceto any specific math skills), self-concept would be the more appropriate term. Choi(2005) claims that specific measures of self-efficacy should better predict performancemore than would general measures of self-concept.

1.1.4 Process versus product criterion variables

One area in particular need of additional study is the ability of self-efficacy to pre-dict process versus product variables. Present research has focused largely on productvariables (such as exam scores, GPA, and course grades). Although Bandura (1977)made no claim that self-efficacy predicts outcome products, self-efficacy may pre-dict process behaviors that lead to particular outcomes. Practicing a particular skilldoes not ensure that one will eventually achieve a desired outcome, but the likeli-hood of the desired achievement is better with practice than without practice. Plus,the belief that one can acquire a skill may affect the degree to which one practices theskill. Consequently, clarifying self-efficacy’s role in contributing to process variables(e.g., studying or engaging in class discussion) may help account for the relationshipbetween self-efficacy and academic outcomes.

Process variables that appear particularly important are procrastination, class atten-dance, on-task behavior, and engagement in class discussion (Klassen et al. 2008).Important product variables include exam performance, course grades, and GPA(Edman and Brazil 2009; Hackett and Betz 1989). An obvious problem with try-ing to predict academic performance from self-efficacy is that academic performance

123

Page 4: The relationship of academic self-efficacy to class participation and exam performance

236 C. E. Galyon et al.

is not one unitary task but a complex sequence of interrelated tasks, such as attend-ing class, taking notes, memorizing specific information, and studying for exams(Wood and Locke 1987). Given the importance of specificity in predicting performancefrom self-efficacy (Choi 2005), one’s attempts to predict academic performance fromgeneral self-efficacy are likely to be minimally successful.

1.2 Self-efficacy and success in college

In attempting to get a sense of the overall relationship between self-efficacy beliefsand academic performance at the college level, we first looked for research syntheses(most particularly meta-analyses) targeting college samples. An early meta-analysisassessed the relationship between self-efficacy and academic outcomes from elemen-tary school through college (Multon et al. 1991). The data base used in the Multon et al.(1991) meta-analysis included 39 samples, with only 11 of those samples comprisedof college students. The average effect size of the college samples (.35) fell within therange for moderate effect sizes according to Cohen’s (1988) criteria, whereas the aver-age effect size for the elementary sample (.21) fell within the range for small effectsizes. The authors of this meta-analysis proposed that this increased effect size forcollege samples might be due to the greater school experience of the college students.

One recent meta-analysis located 109 studies substantiating the relationshipbetween self-efficacy and academic outcomes as reflected by college GPA (Robbins2004). Although standardized test scores (e.g., ACT and SAT) and high school GPAwere consistently the strongest predictors of college GPA, self-efficacy did provesignificantly correlated with college GPA. These authors also conducted regressionanalyses and found that the traditional predictors of college GPA (i.e., SES, highschool GPA, and standardized test scores) better predicted college GPA (R2 = 0.219)

than psychosocial factors (i.e., achievement motivation, academic goals, and academicself-efficacy) (R2 = 0.164). When both sets of predictors were combined, however,significantly more variance in college GPA was explained (R2 = 0.262), with aca-demic self-efficacy proving to be the second strongest contributing factor (B = 0.200)

just behind standardized test scores (R2 = 0.231).In the time frame when the current article was written, the research on self-efficacy

at the college level had become extensive. Our literature search via PsycInfo identified1,652 articles and 1,078 dissertations that included both “self-efficacy” and “college”as keywords. Adding “academic” as a keyword to the “self-efficacy and college” com-bination caused the count to diminish to 344 articles and 370 dissertations—a markedreduction but still a compelling amount of research. One might assume that this vol-ume of research would have established self-efficacy as one of the most powerfuland consistent predictors of success in college. Nonetheless, the most recent researchon the status of self-efficacy as a predictor of success in college has been somewhatmixed. In a word, self-efficacy is a predictor of some measures of college success butnot of other measures, and self-efficacy predicts well under some circumstances butnot under other circumstances (Gore 2010).

Specifically, Gore (2010) reported that self-efficacy measured at the beginning ofthe first year in college is a relatively weak predictor of academic success, predicting

123

Page 5: The relationship of academic self-efficacy to class participation and exam performance

Academic self-efficacy 237

GPA in the range of .00–.20 correlations. On the other hand, when self-efficacy wasmeasured at the end of students’ first semester, correlations with GPA ranged from.35 in the second semester to .21 in the third semester. Not surprisingly, self-efficacybeliefs became stronger predictors of college performance as students progressedthrough college. This increment in predictability may result from students’ seeing thevalue of persistence in achieving success in courses, leading to the belief that diligentand persistent study will produce success in most courses. Thus, students become moreinclined to believe that they can be successful in college courses if they work hard.

Hsieh et al. (2007) examined the relationship between self-efficacy and academicstanding in college. They compared the self-efficacy of two college groups differing intheir academic standing: good academic standing (GPA of 2.0 or higher) and probation-ary academic standing (GPA of less than 2.0). Students in the former group reportedhigher self-efficacy and more mastery goals (developing ability) toward learning thanstudents in the latter group, who reported more performance-avoidance goals (hid-ing lack of ability). Overall, self-efficacy correlated .36 with GPA, .40 with masterygoal orientation, and −.35 with performance–avoidance goal orientation. Nonetheless,some students who reported high self-efficacy were on academic probation. These stu-dents reported significantly more performance-avoidance goals than high self-efficacystudents in good standing.

1.3 Grade point average as a moderator variable

Although grade point average (GPA) is both moderately related to self-efficacy and awidely accepted indicator of general academic performance (Daniels et al. 2009), thebehaviors contributing to GPA need more precise delineation. Academically-orientedbehaviors reflecting high interest in schoolwork, effort to earn high grades, and activeattempts to independently master subject matter are likely contributors to GPA (Cornoand Mandinach 1983; Pressley et al. 1987; Sivan 1986; Zimmerman and Schunk 1989).Academic outcomes, as measured by GPA, are significantly related to autonomouslearning strategies and complex thinking (Pizzolato et al. 2009). Additionally, GPAhas been related to students’ personal development, impulse control, and other positivecharacteristics (e.g., organization, responsibility, and initiative) (Demoulin and Walsh2002; Kirby et al. 2005; Zimmerman and Schunk 1989).

More specifically, GPA has been found to be significantly related to and descrip-tive of student engagement in behaviors leading to academic success, such as con-sistently attending class, participating in class discussion, completing assignments ontime, tracking one’s performance, and contacting instructors outside of class (Bender1997). GPA has also been associated with the strategic learning abilities measured onthe Learning and Study Strategies Inventory (LASSI), including time management,concentration, test strategies, selection of main ideas, self-assessment, and study aids(Everson et al. 2000). Given that GPA represents the relative success of past academicbehaviors, which in turn should affect academic self-efficacy (Lent et al. 1991), stu-dent openness to explicit performance feedback may result in better study strategies,thus contributing to higher GPAs and academic self-efficacy. That being the case, self-efficacy may have a stronger relationship with academic performance for high-GPA

123

Page 6: The relationship of academic self-efficacy to class participation and exam performance

238 C. E. Galyon et al.

students (presumably students who have made good use of performance feedback)than for low-GPA students (who may have been resistant to performance feedback).Thus, we evaluated the possibility of prior GPA serving as a moderator of the relation-ship between current academic self-efficacy and academic performance as reflectedin class participation and exam scores.

2 Framework for the current study

The purpose of the current study was to assess and compare the potential of academicself-efficacy to predict engagement in class discussion (considered a process measure)and scores on major exams (considered an outcome measure), as well as determiningwhether GPA moderates the relationship between self-efficacy and academic variables.On the one hand, we assumed that participating in class is more effort than skill-based,given that students were not graded on the basis of the quality of their comments.In fact, class participation often involves students asking questions to clarify informa-tion rather than demonstrating knowledge of academic content. Howard and Henney(1998) reported that more than 50% of students participate in class discussion primar-ily to clarify information in a course. On the other hand, exam performance is likelybased more on cognitive mastery than mere effort (students received credit based onhow many questions they answered correctly, not how many hours they studied forthe exam or how much time they devoted to taking the exam).

Therefore, we expected self-efficacy to better predict participation (a process behav-ior) than exam performance (an outcome behavior). However, given that students withhigh GPA are generally aware of what behaviors contribute to high course perfor-mance, we expected self-efficacy to better predict exam performance at a higher- thanlower-level of GPA. We did not expect GPA to moderate the relationship betweenself-efficacy and class participation.

3 Method

3.1 Participants

Participants for the study were 165 (37 male, 128 female) undergraduate studentsenrolled in three Educational Psychology course sections of approximately equal size.Demographically, 14% of the participants were freshmen, 42% were sophomores, 22%were juniors, 7% were seniors, and 2% were graduate students. The additional 13%of participants did not specify their academic classification.

3.2 Instrument

The survey used in this study was an adaptation of a measure of academic self-efficacyoriginally developed by Wood and Locke (1987) and later modified by Choi (2005).This independently validated survey consists of seven Likert scale items (1 = stronglydisagree, 5 = strongly agree) that assess course-specific self-efficacy. Each item mea-sures a different task area related to the course, including concentration in class,

123

Page 7: The relationship of academic self-efficacy to class participation and exam performance

Academic self-efficacy 239

memorization, focus on exam, understanding of class concepts, expansion of con-cepts, concept discrimination, and note-taking. An example of an item included onthe survey is, “I can explain the facts, concepts and arguments covered in this courseto others in my own words.” The items on the survey perhaps relate most strongly toexam taking, but are also expected to relate to participation in class discussion (inas-much as the ability to “explain the facts, concepts and arguments covered in this courseto others in my own words” enables participation in class discussion). Self-efficacyscores could range from 7 to 35. Computed reliability coefficients across all units andall sections yielded an average alpha coefficient of 0.92.

3.3 Data collection

At the beginning of the course, students were asked to participate in a research studyand fill out a demographic survey that requested their previous GPA and their expectedgrade in the course. As demonstrated in other studies (e.g., Cassady 2001), self-reported GPA is a reasonably accurate measure of student GPA and consequentlyappropriate for use in research. Immediately prior to each of five major 50-itemmultiple-choice exams, students were asked to fill out the self-efficacy survey andsubmit the completed form to the instructor prior to the exam. Students recorded theirexam answers on scan sheets. Following the completion of the exam, a test scanningmachine generated an exam score for each student and identified the correct answers toitems missed by each student. This arrangement permitted immediate individual examfeedback following student completion of the exam. Feedback included the total scoreon the exam and the correct answer to each item missed. Self-efficacy ratings werematched to student exam scores via student names. The names were then matched toa research number and entered into a database.

Participation data were collected via classroom observation on the four discussiondays in each of the five units, making a total of 20 days of observation. Observers(graduate teaching assistants) recorded the number of voluntary comments made byeach student. Voluntary comments were characterized as unsolicited responses toinstructor and classmate questions, as well as questions posed by the student. Neitherquestions unrelated to content issues nor quality of student comments was includedin the record of participation. One observer recorded participation each discussionday and another recorded participation one day per unit. Inter-rater agreement on theinter-rater check day was determined by dividing the smaller number of commentsrecorded by one observer by the larger number of comments recorded by the otherobserver (97.69% average inter-rater agreement across units and course sections).In some units of each course section, a small amount of credit was given for one to twocomments in class discussion. The units and days within those units when credit wasawarded for participation were randomly determined at the conclusion of the course.

4 Results

In determining the linkage between self-efficacy scores and the academic variables,we first correlated self-efficacy with participation and exam scores for each unit in

123

Page 8: The relationship of academic self-efficacy to class participation and exam performance

240 C. E. Galyon et al.

Table 1 Correlations of self-efficacy with exam performance and participation across units

Exam performance Participation

r p N r p N

Unit A self-efficacy 0.045 0.569 160 0.094 0.239 160

Unit B self-efficacy 0.115 0.156 153 0.080 0.321 155

Unit C self-efficacy 0.234∗∗ 0.003 155 0.107 0.185 155

Unit D self-efficacy 0.243∗∗ 0.004 142 0.093 0.272 140

Unit E self-efficacy 0.194∗ 0.016 154 0.200∗ 0.013 154∗ indicates significance at P < 0.05∗∗ indicates significance at P < 0.01

the course (Table 1). Correlational analyses indicated that self-efficacy was signifi-cantly related to participation in class discussion in only the last unit of the course(r = 0.200, p = 0.013). In contrast to the correlations between self-efficacy andparticipation scores, significant correlations were found between self-efficacy andexam scores for Units C, D, and E, with r = 0.234(p < 0.01), r = 0.243(p <

0.01), and r = 0.194(p < 0.05), respectively across combined course sections. Thus,self-efficacy was more related to exam performance from the mid-point on in the coursethan earlier in the course, and self-efficacy was more related to exam scores than partic-ipation in class discussion. Furthermore self-efficacy scores tended to remain relativelystable throughout the semester with mean scores ranging from M = 26.21 to 27.47.

As a means of comparing relationships between self-efficacy and the academicvariables on a group basis, we used cluster analysis to identify composite self-efficacygroups for combined units in the course. This analysis yielded three levels of self-efficacy, while retaining the maximum portion of the original sample: “Low” —M =22.30, SD = 2.36, Min = 14.00, Max = 25.75, n = 40; “Mid”—M = 27.16, SD = 1.39,Min = 24.25, Max = 31.00, n = 91; “High”—M = 30.58, SD = 1.43, Min = 28.00,Max = 34.50, n = 34.

To determine the relationship between placement in these self-efficacy groups andplacement in extreme academic groups, we used quartiles to identify students withhigh or low participation scores and high or low exam scores. For cumulative com-ments, the low-participation scores ranged from 0 to 0.54 average comments per unit,M = 0.20(n = 39) and the high- participation scores ranged from 2.31 to 7.88 aver-age comments per unit, M = 3.48(n = 39). For percentage of correct responses onexams, the low-exam percentages ranged from 50.75 to 73.20, M = 65.76(n = 43)

and the high-exam percentages ranged from 82.80 to 98.80, M = 88.68(n = 36).Frequencies of group membership were computed for combinations of (a) self-

efficacy groups and participation groups and (b) self-efficacy groups and exam-performance groups. K-independent proportions tests (Ferguson and Takane 1989)were first used to compare the proportionality of low and high participation at eachself-efficacy level. Percentage-wise, students with low self-efficacy were significantly(p < .001) more likely to be low than high participants (Table 2). Likewise, midself-efficacy students were significantly (p < .01) more likely to be low than highparticipants. Conversely, students with high self-efficacy were more likely to be

123

Page 9: The relationship of academic self-efficacy to class participation and exam performance

Academic self-efficacy 241

Table 2 Self-efficacy and class participation group membership

Self-efficacy groupa Participation groupb N %

Low SE Low participation 11 61.1

High participation 7 38.9

Mid SE Low participation 20 58.8

High participation 14 41.2

High SE Low participation 8 30.8

High participation 18 69.2a Self-efficacy groups (Low, Middle, and High) were identified via cluster analysis of average self-efficacyscores for the semesterb Participation group selection (Low and High) was based on quartiles of cumulative comments made duringthe semester

Table 3 Self-efficacy and exam performance group membership

Self-efficacy groupa Exam groupb N %

Low SE Low exam 11 61.1

High exam 7 38.9

Mid SE Low exam 24 63.2

High exam 14 36.8

High SE Low exam 8 34.8

High exam 15 65.2a Self-efficacy groups (Low, Middle, and High) were identified via cluster analysis of average self-efficacyscores for the semesterb Exam-group selection (Low and High) was based on quartiles of average exam scores for the semester

high than low participants (p < .001). Thus, only membership in the high self-efficacygroup was significantly related to membership in the high-participation group.Membership in the other self-efficacy groups was significantly linked to placementin the low-participation group.

With regard to placement in the low and high-exam groups, a significantly (p <

.001) higher percentage of students with low self-efficacy were in the low rather thanthe high-exam group (Table 3). Also, a higher percentage of students in the mid self-efficacy group were in the low rather than the high-exam group (p < .001). Only thestudents with high self-efficacy scores were more likely to be in the high-exam thanthe low-exam group (p < .001). Thus, high-exam performance was linked only toa high level of self-efficacy. Students having a mid or low level of self-efficacy weremore likely to perform poorly than well on the exams.

With respect to the linkage between GPA and both self-efficacy and the academicvariables, we first correlated GPA with self-efficacy, participation, and exam scores.The correlation between GPA and exam performance was significant, r = 0.354,

p < .001, but the correlations between GPA and both participation (r = 0.017,

p = .839) and self-efficacy (r = 0.016, p = .814) were not significant. However,these correlations may have been somewhat diminished by the study’s constrained

123

Page 10: The relationship of academic self-efficacy to class participation and exam performance

242 C. E. Galyon et al.

range of GPA scores (due to selection criteria for admissions to the University). Forthat reason, cluster analysis was used to identify statistically distinct levels of GPA:Level 1 (GPA = 3.53–4.00), Level 2 (GPA = 2.98–3.50), Level 3 (GPA = 2.30–2.95),and Level 4 (GPA = 1.30–2.19). A one-way ANOVA confirmed that the levels differedsignificantly on GPA, F(3, 204) = 12.36, p < .001. Post-hoc comparisons revealedthe following pattern of level differences: Level 1 had a higher GPA than all otherlevels (p < .05 or lower); Level 2 significantly differed from Level 1 (p < .05);and Level 3 significantly differed from Levels 1 and 2 (p < .05). Because Levels 3and 4 did not differ significantly from each other, they were combined in subsequentanalyses, resulting in three GPA levels (High, Moderate, and Low).

We then determined whether self-efficacy, participation, and exam means differedby GPA levels. An ANOVA showed no significant differences in levels of self-effi-cacy for high, middle, and low GPA levels, F(2, 204) = 1.652, p = 0.194. Likewise,mean participation level did not differ significantly across GPA levels. Exam scoresdid differ significantly, F(2, 205) = 15.181, p < .001, across GPA levels. Post-hoccomparisons showed significant differences between all GPA groups in exam scores(p < .001 to p = .017), with the High GPA level scoring significantly higher onexams than the Middle and Low GPA levels and the Middle GPA level scoring signif-icantly higher than the Low GPA level.

The relationships between self-efficacy and both academic variables were alsoassessed at the different levels of GPA. Mean scores of all variables were calculatedfor each level of GPA (see Table 4). Self-efficacy scores at the High-GPA level cor-related significantly with both participation (r = 0.452, p = .002) and exam per-formance (r = 0.283, p = .023). Self-efficacy scores at the Middle-GPA level didnot correlate significantly with participation (r = 0.078, p = .561) but did correlate

Table 4 Descriptive statistics for self-efficacy, exam, and participation measures per GPA group

GPA groupa Variableb N Range Mean SD

High GPA 66 0.47 3.75 0.144

Self-efficacy 66 16.25 27.43 3.310

Unit. exams 65 20.50 40.86 4.326

Participation 46 4.50 1.47 1.056

Middle GPA 78 0.52 3.22 0.176

Self-efficacy 77 18.67 26.45 3.548

Unit exams 78 22.25 38.68 4.456

Participation 58 6.33 1.64 1.521

Low GPA 65 1.65 2.57 0.376

Self-efficacy 64 20.50 27.26 3.523

Unit exam 65 19.25 36.61 4.407

Participation 42 7.88 1.68 10.142a GPA groups were identified via cluster analysis of student-reported GPAs. Means, SD, and ranges werecalculated for each groupb Mean self-efficacy, unit exam scores and participation were computed across the entire semester

123

Page 11: The relationship of academic self-efficacy to class participation and exam performance

Academic self-efficacy 243

Table 5 Correlations of self-efficacy, exam, and participation measures per GPA group

GPA groupa Variableb Self-efficacyr p N

High Participation 0.452 0.002** 46

Unit Exams 0.283 0.023* 65

Middle Participation 0.078 0.561 58

Unit Exams 0.233 0.040* 78

Low Participation 0.100 0.531 42

Unit Exams 0.133 0.290 65a GPA groups were identified via cluster analysis of student-reported GPAs. Means, Standard deviations,and ranges were calculated for each groupb Mean self-efficacy, unit exam scores and participation were computed across the entire semester* indicates significance at p < 0.05** indicates significance at p < 0.01

significantly with exam performance (r = 0.233, p = .040). Lastly, students at theLow-GPA level produced no significant relationships between self-efficacy and eitherparticipation (r = 0.100, p = .531) or exam performance (r = 0.133, p = .290).Thus, self-efficacy was more likely to correlate significantly with both participationand exam performance at the high GPA level than at the mid and lower GPA levels,and was more likely to correlate significantly with exam performance at the mid GPAlevel than at the lower GPA level (see Table 5).

5 Discussion

This study shows that certain levels of academic self-efficacy may be predictive ofparticipation in class discussion and exam performance. Inasmuch as students in allsections of the targeted course received some credit for participating in discussion,irrespective of the quality of comments, participation could be considered a measureof course engagement rather than performance level. On the other hand, exam gradeswere based on the number of items answered correctly, rather than on how much timestudents invested in studying for the exam or taking the exam. Consequently, gradeson an exam could be considered more of an outcome measure than a task-engagementmeasure. Given previously reported relationships of self-efficacy to task engagementand performance, one might have expected stronger relationships between self-effi-cacy and participation than between self-efficacy and exam performance (Bandura1997; Pajares 2003).

5.1 Levels of self-efficacy

Cluster analysis showed that self-efficacy was predictive of participation and examperformance at all three levels of self-efficacy. However, the directionality of thoserelations differed across the self-efficacy levels. Students at the high self-efficacy level

123

Page 12: The relationship of academic self-efficacy to class participation and exam performance

244 C. E. Galyon et al.

were more likely to be in the high than the low-participation group. Conversely, stu-dents at both the mid and low self-efficacy levels were more likely to be in the lowthan high-participation groups. Exactly the same pattern held for self-efficacy levelsand exam groups. Consequently, as indicated previously (see pages 12 and 13, andTables 2 and 3), self-efficacy may constitute an impetus for academic engagement andperformance only at the highest level of self-efficacy.

A second major finding of the study was that GPA appeared to partly moderate therelationship between self-efficacy and exam performance. The same variables that mayelevate GPA (e.g., perseverance in study habits and openness to performance feed-back) likely also elevate exam performance. Low perseverance in study habits andlack of responsiveness to performance feedback likely contribute to both low GPAsand low exam performance, irrespective of self-efficacy scores. Surprisingly, the meanlevels of self-efficacy were virtually the same at the high, middle, and low GPA levels.This pattern may suggest that low and middle GPA students are relatively imperviousto how well or how poorly prepared they are for exams. Another possibility is that thelow GPA students are inclined to overestimate their preparation for exams, perhaps inpart to present a favorable image of their ability.

Insofar as major exams are often regarded as primary measures of student learningin college courses, instructors should consider methods for helping students realis-tically judge how well prepared they are for exams. Overall, the low GPA studentsregarded themselves as academically skilled (as represented by self-efficacy) as thehigh GPA group. Also, student judgments of their self-efficacy were less predictive oftheir exam performance at the low GPA levels than at the high GPA level. Studentsmay benefit from instructors identifying precise preparation strategies for studentsto use in judging their preparation for exams. In the course in which this study wasconducted, instructors have recently given very specific suggestions as to how bestto prepare for exams. We are now beginning to monitor the extent to which studentsactually use these strategies and the relationship between use of these strategies andexam scores.

Although the current study did not attempt to elevate academic self-efficacy, thefindings regarding the percentages of students in the high and low participation andexam groups suggest that only elevating self-efficacy to its highest level (e.g., viatask-analysis and explicit performance feedback) will make a positive difference inacademic measures. Elevating self-efficacy to a moderate level is unlikely to lead tohigher class participation or exam scores. In fact, both low and moderate academicself-efficacy in the current study predicted a higher percentage of low participation andlow exam scores than high participation and high exam scores (see Tables 2 and 3).

Rather than representing self-efficacy’s predictive potential on a continuum fromlow to high, academic self-efficacy appears to have dichotomous predictive potential.Students must strongly believe they can perform an academic task—not just mod-erately believe they can perform that task—for their belief to make a positive differ-ence in academic measures. Mid-level academic self-efficacy may reflect ambivalenceabout one’s potential to participate in class discussion or score high on course exams.Any doubt that one can perform a target behavior may be as limiting as having littleconfidence that one can perform that behavior. Uncertainty about one’s ability to per-form a particular skill may create hesitancy to attempt the skill or a tendency to err

123

Page 13: The relationship of academic self-efficacy to class participation and exam performance

Academic self-efficacy 245

in attempts to perform the skill. In between the self-statements that “I can’t do it”and “I can do it” is the statement of “Maybe I can do it.” But the reciprocal of thelatter statement is “Maybe I can’t do it.” That element of doubt is what may havepre-empted a positive impact of mid-level self-efficacy on participation and examperformance.

5.2 Limitations of the study

A principal limitation of the current study is the weak to moderate strength of thefindings. Although many of the correlations and cluster analyses yielded statisticallysignificant results, they left much of the variance and exceptions to general trends unex-plained. For example, correlations in the .20 to .30 range may have been statisticallysignificant, but these correlations would still be considered weak according to Cohen’s(1988) criteria. Despite the finding that self-efficacy levels predicted membership inhigh and low participation groups, as well as membership in high and low exam-scoregroups, a sizeable percentage of students deviated from those significant trends. Forexample, nearly 40% of students in the low self-efficacy group found placement in thehigh participation and/or the high exam group. On the other hand, a little over 30%of the high self-efficacy group found placement in the low participation and/or lowexam group. So while group memberships differed significantly across self-efficacylevels, these membership trends were characterized by numerous exceptions. Obvi-ously, variables above and beyond self-efficacy were contributing to participation andexam performance at all levels of self-efficacy.

Another limitation was the relatively small sample size. This problem was fur-ther exacerbated by the reduction in sample size due to grouping at three levels onself-efficacy and two levels on each academic variable, which produced very smallmemberships in the comparison groups. Not surprisingly, the greatest reduction inns occurred in the high and low self-efficacy groups rather than in the mid self-effi-cacy group. In some respects, the group ns suggest the possibility of a normal curvefor self-efficacy scores in the sample. Nonetheless, the results obtained for the middleself-efficacy group extended the pattern found in the low self-efficacy group. The mostfavorable support for the potency of self-efficacy occurred in the high self-efficacygroup. It was in this group that self-efficacy appeared to have the most positive effecton participation and exam performance. However, the generalizability of that patternis tempered by the fact that only 18 students represented the high self-efficacy/highparticipation combination and only 15 students represented the high self-efficacy/highexam combination.

The measure of self-efficacy in the current study also focused on a variety ofacademically-related behaviors (e.g., note-taking, concept comprehension, conceptrecollection, and ability to focus) rather than exam-specific behaviors. As mentionedpreviously, other research has found that more specific measures of self-efficacybetter predict outcomes than more general ones (Choi 2005). A survey that specif-ically assesses self-efficacy regarding exam performance may better predict examoutcomes than does a more general academic self-efficacy measure. Additionally,inasmuch as self-efficacy is a product of prior experience and GPA ostensibly reflects

123

Page 14: The relationship of academic self-efficacy to class participation and exam performance

246 C. E. Galyon et al.

that same experience, a measure of general academic self-efficacy at the beginning of acourse may improve our understanding of how self-efficacy influences student behav-iors (e.g., participation, note-taking, and study habits) and leads to favorable examoutcomes.

Although the study was not designed to increase self-efficacy, that remains a veryimportant challenge. In one sense, the findings of the current study have raised the barof that challenge. Increasing self-efficacy from a low to moderate level may have littleimpact on the academic measures to which self-efficacy is conceptually linked. Ourfindings suggest that getting students to a level where they believe unequivocally thatthey can perform a certain skill or at least learn to perform that skill may be necessaryto see positive changes in the academic performance of those students.

5.3 Contributions to the self-efficacy literature

Despite the small samples in the group analyses, this study raises a possibility thathas received scant attention in the self-efficacy literature. That issue is whether self-efficacy must be very high before it becomes a contributor to academic performance.Minimal or even modest increases in self-efficacy are not likely to be paralleled byminimal or modest increases in academic variables. Our relatively weak correlationsbetween self-efficacy and academic measures raise doubts as to the robustness ofoverall relationships between self-efficacy and academic measures. Only when self-efficacy is at a very high level is it likely to override the impact of other variablescontributing to academic performance.

This emphasis on the exclusive potency of high self-efficacy is counter to someprevious claims in the literature. Multon et al. (1991) argued in their meta-analysisthat self-efficacy can be too high, leading to gross overestimations of one’s abilityand causing one to attempt tasks well beyond one’s potential. Plus, high self-effi-cacy can cause one to be complacent in preparing for a task (Vancouver and Kendall2006). Several studies conducted by Vancouver et al. (2001, 2002) have reported self-efficacy to be negatively related to performance, presumably because of diminishedeffort. According to Vancouver and Kendall, low self-efficacy may be more likely toencourage one to work harder than would high self-efficacy. But as pointed out earlierin our article, the potency of high self-efficacy may differ depending on whether weare referring to perception of current skill or the prospect of developing a particularskill.

Methods designed to promote self-efficacy have generally produced positive results.These methods include providing feedback and encouragement, allowing for vicari-ous observation of similar peers overcoming obstacles to success, and promoting mas-tery of information related to an academic domain (Schunk and Meece 2006; Usherand Pajares 2008). Specifically, Jackson (2002) reported self-efficacy enhancementeffects after teachers provided notes of encouragement and feedback to individual stu-dents following poor test performance. Susskind (2003) found that students reportedhigher academic self-efficacy regarding their note-taking and studying techniques afterattending traditional lectures that included posted organized notes than students whoattended traditional lectures without this supplement. Luzzo et al. (1999) noted higher

123

Page 15: The relationship of academic self-efficacy to class participation and exam performance

Academic self-efficacy 247

math and science self-efficacy in undeclared college students who participated in anintervention that promoted mastery of information under time pressure compared tostudents who received an intervention that involved vicarious learning of self-efficacy.Not all methods targeting self-efficacy, however, have been successful. For example, aself-correction strategy designed to promote self-efficacy and motivation in a middleschool mathematics class yielded little to no change in self-efficacy (Ramdass andZimmerman 2008).

Although self-efficacy is a conceptually appealing notion, its potential as a practicalpredictor of tangible performance measures may be circumscribed. Its linkages to per-formance have been modest at best and sometimes inconsistent. Even the desirabilityof promoting self-efficacy has been called into question. Perhaps the safest route topromoting self-efficacy is developing skills to which self-efficacy would presumablybe linked. As noted by Bandura (1994) and Vancouver and Kendall (2006), strongerself-efficacy may be a product of skill development. For students weak in target skills,breaking down these skills into very small steps (task analysis) that could be success-fully executed by students may improve self-efficacy and lead to skill development.The acid test of self-efficacy is whether high self-efficacy gets one to a higher skilllevel faster than does low or moderate self-efficacy. But after a high skill level has beenattained, high self-efficacy will not necessarily predict how efficiently and effectivelyone will use skills at that level.

References

Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. PsychologyReview, 84, 191–215.

Bandura, A. (1994). Self-efficacy. In V. S. Ramachandran (Ed.), Encyclopedia of human behavior(Vol. 4, pp. 71–81). New York: Academic Press. (Reprinted in H. Friedman [Ed.], Encyclopediaof mental health. Sand Diego: Academic Press, 1998).

Bandura, A. (1997). Self-efficacy: The exercise of control. New York: W. H. Freeman.Bender, D. S. (1997). Effects of study skills programs on the academic behaviors of college students.

Paper presented at the Annual Meeting of the Eastern Educational Research Association.Cassady, J. C. (2001). Self-reported GPA and SAT scores. ERIC Digest. Retrieved from ERIC database.

(ED458216).Choi, N. (2005). Self-efficacy and self-concept as predictors of college students’ academic perfor-

mance. Psychology in the Schools, 42, 197–205.Chowdhury, M. S., & Shahabuddin, A. M. (2007). Self-efficacy, motivation and their relationship to

academic performance of Bangladesh college students. College Quarterly, 10, 1–9.Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence

Erlbaum Associates, Inc..Corno, L., & Mandinach, E. B. (1983). The role of cognitive engagement in classroom learning and

motivation. Educational Psychologist, 18, 88–108.Daniels, L. M., Stupnisky, R. H., Pekrun, R., Haynes, T. L., Perry, R. P., & Newall, N. E. (2009).

A longitudinal analysis of achievement goals: From affective antecedents to emotional effects andachievement outcomes. Journal of Educational Psychology, 101, 948–963.

Demoulin, D. F., & Walsh, R. J. (2002). Comparing the personal development test scores against GPA,grade, and age of typical high school students. Journal of Instructional Psychology, 29, 22–24.

Edman, J. L., & Brazil, B. (2009). Perceptions of campus climate, academic efficacy and academicsuccess among community college students: An ethnic comparison. Social Psychology of Education:An International Journal, 12, 379–383.

123

Page 16: The relationship of academic self-efficacy to class participation and exam performance

248 C. E. Galyon et al.

Everson, H. T., Weinstein, C. E., & Laitusis, V. (2000). Strategic learning abilities as a predictorof academic achievement. Paper presented at the Annual Meeting of the American EducationalResearch Association, New Orleans, LA.

Ferguson, G. A., & Takane, Y. (1989). Statistical analysis in psychology and education (6th ed.).New York: McGraw-Hill Publishing.

Finney, S. J., & Schraw, G. (2003). Self-efficacy beliefs in college statistics courses. ContemporaryEducational Psychology, 28, 161–186.

Gore, P. A. (2010). Academic self-efficacy as a predictor of college outcomes: Two incremental validitystudies. Journal of Career Assessment, 14, 92–115.

Hackett, G., & Betz, N. E. (1989). An exploration of the mathematics self-efficacy/ mathematicsperformance correspondence. Journal of Research in Mathematics Education, 20, 263–271.

Howard, J. R., & Henney, A. L. (1998). Student participation and instructor gender in the mixed-agecollege classroom. Journal of Higher Education, 69, 384–405.

Hsieh, P., Sullivan, J. R., & Guerra, N. S. (2007). A closer look at college students: Self-efficacy andgoal orientation. Journal of Advanced Academics, 18, 454–476.

Jackson, J. W. (2002). Enhancing self-efficacy and learning performance. The Journal of ExperimentalEducation, 70, 243–254.

Kirby, K. N., Winston, G. C., & Santiesteban, M. (2005). Impatience and grades: Delay-discount ratescorrelate negatively with college GPA. Learning and Individual Differences, 15, 213–222.

Klassen, R. M., Krawchuk, L. L., & Rajani, S. (2008). Academic procrastination of undergraduates: Lowself-efficacy to self-regulate predicts higher levels of procrastination. Contemporary EducationalPsychology, 33, 915–931.

Lent, R. W., Brown, S. D., & Gore, P. A. (1997). Discriminant and predictive validity of academicself-concept, academic self-efficacy, and mathematics-specific self-efficacy. Journal of CounselingPsychology, 44, 307–315.

Lent, R. W., Lopez, F. G., & Bieschke, K. J. (1991). Mathematics self-efficacy sources and relation toscience-based career choice. Journal of Counseling Psychology, 38, 424–430.

Luzzo, D. A., Hasper, P., Albert, K. A., Bibby, M. A., & Martinelli, E. A. (1999). Effects of self-efficacyenhancing interventions on the math/science self-efficacy and career interests, goals, and actionsof career undecided college students. Journal of Counseling Psychology, 46, 233–243.

McCombs, B. L., & Marzano, R. J. (1990). Putting the self in self-regulated learning: The self as agentin integrating will and skill. Educational Psychologist, 25, 51–69.

Multon, K. D., Brown, S. D., & Lent, R. W. (1991). Relation of self-efficacy beliefs to academicoutcomes: A meta-analytic investigation. Journal of Counseling Psychology, 38, 30–38.

Owen, S. V., & Froman, R. D. (1988). Development of a college academic self-efficacy scale.Paper presented at the Annual Meeting of the National Council on Measurement in Education,New Orleans, LA.

Pajares, F. (2003). Self-efficacy beliefs, motivation, and achievement in writing: A review of theliterature. Reading and Writing Quarterly, 19, 139–158.

Pizzolato, J. E., Hicklen, S. T., Brown, E. L., & Chaudhari, P. (2009). Student development, studentlearning: Examining the relation between epistemologic development and learning. Journal ofCollege Student Development, 50, 475–490.

Pressley, M., Borkowski, J. G., & Schneider, W. (1987). Cognitive strategies: Good strategy userscoordinate metacognition and knowledge. Annals of Child Development, 4, 89–129.

Ramdass, D., & Zimmerman, B. J. (2008). Effects of self-correction strategy training on middle schoolstudents’ self-efficacy, self-evaluation, and mathematics division learning. Journal of AdvancedAcademics, 20, 18–41.

Robbins, S. B., Huy, L., & Davis, D. (2004). Do psychosocial and study skill factors predict collegeoutcomes? A meta-analysis. Psychological Bulletin, 130, 261–288.

Schunk, D. H., & Meece, J. L. (2006). Self-efficacy in adolescence. In F. Pajares & T. Urdan (Eds.),Adolescence and education (Vol. 5, pp. 71–96). Greenwich, CT: Information Age Publishing.

Sivan, E. (1986). Motivation in social constructivist theory. Educational Psychologist, 21, 209–233.Susskind, J. E. (2003). PowerPoint’s power in the classroom: Enhancing students’ self-efficacy and

attitudes. Computers & Education, 45, 203–215.Usher, E. L., & Pajares, F. (2008). Sources of self-efficacy in school: Critical review of the literature

and future directions. Review of Educational Research, 78, 752–796.

123

Page 17: The relationship of academic self-efficacy to class participation and exam performance

Academic self-efficacy 249

Vancouver, J. B., & Kendall, L. N. (2006). When self-efficacy negatively relates to motivation andperformance in a learning context. Journal of Applied Psychology, 91, 1146–1153.

Vancouver, J. B., Thompson, C. M., Tischner, E. C., & Putka, D. (2002). Two studies examin-ing the negative effect of self-efficacy on performance. Journal of Applied Psychology, 87,505–516.

Vancouver, J. B., Thompson, C. M., & Williams, A. A. (2001). The changing signs in the relation-ship among self-efficacy, personal goals, and performance. Journal of Applied Psychology, 86,605–620.

Wood, R. E., & Locke, E. A. (1987). The relation of self-efficacy and grade goals to academicperformance. Educational and Psychological Measurement, 47, 1013–1024.

Zimmerman, B. J., & Schunk, D. H. (1989). Self-regulated learning and academic achievement: Theory,research, and practice. New York: Springer.

Author Biographies

Charles E. Galyon is currently a fourth year doctoral student in the School Psychology program at theUniversity of Tennessee, Knoxville. His research interests include identifying contributors to student suc-cess, improving student success through instructional design, and academic interventions in early education.

Carolyn A. Blondin is a fourth-year doctoral student in the School Psychology program at the Universityof Tennessee, Knoxville. She earned a B.A. in Psychology from Elon University in 2006 and received herM.S. in Educational Psychology from the University of Tennessee in 2010. She is currently working on herdissertation, which focuses on early academic interventions at the college level.

Jared S. Yaw is a fourth-year doctoral student in the School Psychology program at the University ofTennessee, Knoxville. Jared has served as a teaching assistant and instructor in a large, undergraduateeducational psychology course and career exploration counselor education course. His research interestsinclude working with college students to increase participation and self-efficacy, as well as improvingsight-word reading among school aged children with disabilities.

Meagan L. Nalls is currently a fifth-year doctoral student in the School Psychology program at theUniversity of Tennessee, Knoxville. Her research interests include academic interventions, parent involve-ment in education, and academic motivation.

Robert L. Williams is a Professor of School Psychology at the University of Tennessee, Knoxville. He hasco-authored five books and published well over 100 articles in professional journals. His research topicsinclude college students’ critical thinking, writing skills, participation in class discussion, sociopoliticalvalues, and stress levels.

123