The accuracy of self-efficacy: a comparison of high school and college students.

Abstract

There are perhaps two views of what constitutes accurate self-efficacy: the competence view (based on norm-referencing) and the beliefs view (based on criterion-referencing). In this study, we compared the accuracy of self-efficacy for a punctuation task across three educational levels (11th grade students, college freshmen, and college juniors) and found differences in accuracy among the groups from both the competence and the beliefs views. Results are attributed to the types of correctional feedback teachers give at each educational level.

**********

Students who are confident in their ability to be successful on a specific task are likely to be motivated to engage in the task (Bandura, 1997). Since its inception, this premise, known as self-efficacy, has driven lines of motivational research that have deepened our understanding of the relationship between confidence and performance. Unfortunately, little empirical work has been done to broaden our understanding of students' self-efficacies in terms of accuracy (Pajares, 1996; Pintrich & Schunk, 1996). In this study we investigate the accuracy of self-efficacy among educational levels (11th grade students, college freshmen, and college juniors) for a specific punctuation task. We then analyze the degree to which efficacy accuracy (EA) can vary within and among the groups and attribute the differences in EA to the types of corrective feedback given by teachers at each level of education.

Efficacy Differences

A person may have different levels of self-efficacy for various tasks depending upon the nature of each task and a person's experiences (Bandura, 1997). Similarly, self-efficacy for a task can differ among people. Factors such as modeling, verbal persuasion, and psychological state, can also affect people's self-efficacy and, in turn, influence motivation, task orientation, effort, persistence, beliefs, strategy use, and performance (Bandura, 1997; Schunk, 1994).

There is general agreement on the definition of self-efficacy and its effects on other learning factors, but there are perhaps two views of what constitutes accurate self-efficacy. From the competence view, efficacy is accurate for a group when individuals with the highest self-efficacy perform the best and individuals with the lowest self-efficacy perform the worst (a quasi-norm-referenced view) (Multon et. al., 1991; Pintrich & Schunk, 1996; Ryan & Deci, 1998; Shell et al, 1995). However, norm-referencing may not tell us the entire story. From the beliefs view, an individual's self-efficacy is viewed separate from the group and may, at some level, be overestimated or discounted when performance is considered in reference to a course goal (a quasi-criterion-referenced view) (Bandura, 1997; Pajares, 1996; Pajares et al, 1999; Pintrich & Schunk, 1996). In other words, high self-efficacy may not necessarily predict a high level of competence (A+ work), and low self-efficacy may not predict a low level of competence (D work) outside of one's peer group. Though much evidence has been found in support of the competence view of EA, limited research has addressed the beliefs view of EA. In general, results have suggested there are differences in EA between genders and among ability levels (Ewers & Wood, 1993; Glenberg & Epstein 1985) and among developmental levels (Lachman & Jelalian, 1984; Kaley & Cloutier, 1984; Pressley, Levin, Ghatala, & Ahmad, 1987), but those beliefs are subject to change (Meier, McCarthy, & Schmeck, 1984; Stone, 1994; Weaver, 1990).

The purpose of the present study is to examine, from both the beliefs view and the competence view, EA for punctuating sentences. Punctuation was selected as the learning task for three reasons: (1) because little has been done to investigate self-efficacy for writing tasks (Pajares et al, 1999), (2) existing research on self-efficacy for writing tasks is mixed (Meier, et. al., 1984; Shell et al., 1995), and (3) writing tasks address a general skill of which punctuation is a specific sub-skill; specific skills yield the most accurate results in self-efficacy studies (Bandura, 1997).

Our choice of participant groups was also a function of previous research (Meier et al., 1984, Shell et al, 1995). In a study by Meier et al (1984), the efficacy of college freshmen was predictive of performance at the beginning of a writing course, but it was not predictive at the end of the course. We were curious to see if we could replicate their results with a more specific writing task. In addition, we sought to document and analyze any trends that might appear in educational levels before and after that seemingly critical period (freshman year of college). Shell et al (1995) documented no differences in student beliefs regarding writing tasks between the 3rd and 10th grades. We chose to continue the developmental examination where they left off. Each educational group was separated by two educational years--a design decision we thought would sufficiently display variability among groups, if variability indeed existed. It is important that the EA of these three groups be addressed. From a practical standpoint, students entering college are expected to be independent learners (Schunk & Zimmerman, 1998); they should have accurate perceptions of what they are capable of accomplishing on certain academic tasks in order to enroll in appropriate courses or plan for remediation. It is critical that teachers be aware of the inaccuracies of their students' confidences as well, so that they can give students the proper corrective feedback to promote learning, maintain motivation (Bandura, 1997; Pajares et al, 1999; Pintrich & Schunk, 1996), and help students become self-regulated learners (Schunk & Zimmerman, 1998).

Methods and Data Source

Participants in this study included 11th grade students from a Midwestern high school and freshmen and juniors from a large, Midwestern university. Forty percent of the total sample (n = 144) were college juniors, 32% were high school juniors, and 28% were college freshmen. Fifty-three percent of the sample was female, while 47% was male. The majority of the families serviced by the schools in the area were of middle class socio-economic status, and the majority of the students were White. All students volunteered to participate.

Instruments

Punctuation Self-Efficacy. There was a total of 11 items. Nine items on the efficacy for punctuation measure were selected from a web-based tutorial. Two, additional items were constructed by the researchers and were reviewed for clarity by a grammar expert. The guidelines prescribed by Bandura (1997) were followed in the construction of the efficacy instrument, which contained a 5-point Likert-type scale, ranging from 1 = not at all confident through 5 = absolutely certain.

Punctuation Performance. The 11 items in the punctuation performance measure mirrored those from the self-efficacy measure. Each item was scored to parallel the 5-point value of the self-efficacy Likert-type scale, ranging from 1 = no punctuation correct to 5 = all punctuation marks were correct.

Procedures

Each student was given the efficacy measure before the performance measure (Bandura, 1997). Students were given explanations of a 5-point Likert-type scale and were told that each item contained one sentence. For the performance measure, the students were instructed to complete only punctuation. Performances were assessed in accordance with the grammar rules that are prescribed by the school district and the university in which our participants were students.

Data Analysis and Results

Differences in Beliefs and Performances. A one-way analysis of variance revealed that self-efficacy did not differ among the groups. The ANOVA was not statistically significant, F(2, 141) = 2.58, p>.05, []2 = .04. Another one-way ANOVA revealed that performance did differ among the groups. The ANOVA was statistically significant, F(2, 141) = 44.11, p<.001. The strength of the relationship between year in school and punctuation performance, as assessed by eta-squared ([]2), was strong, with the year in school factor accounting for 39% of the variance of punctuation performance. Follow-up tests, using Tukey-HSD, revealed that performance increased with educational level. College juniors (M = 47.31, SD = 4.18) outperformed college freshman (M = 43.32, SD = 5.22), and college freshmen outperformed high school juniors (M = 37.51, SD = 6.41).

Efficacy-Performance Relationship. Fischer's z transformation tests (presented in Table 1) show that correlation coefficients between efficacy and performance for high school juniors, college juniors, and all groups combined were statistically significant (greater than or equal to .42), but for the college freshmen it was not.

Accuracy of Estimating Punctuation Performance. Dependent t-tests indicated that high school juniors and college freshman significantly overestimated their punctuation performance, t(45) = 7.1, p<.001, []2 = .53, and t(40) = 5.1, p<.001, []2 = .4, respectively, whereas college juniors did not. The results of a regression analysis yielded differences in the two models explored. First, the self-efficacy model was found to be significant and had an R2 = .18, (F(1,142)) = 30.8, p<.01. Then, the self-efficacy model with year in school as another predictor was found to also be significant, R2 = .56, (F(1,140)) = 59.92, p<.01 (Table 3). Adding year in school to the self-efficacy model did increase the R-square significantly (R2-change = .24, (F(1,140)) = 63.33, p<.01. See issue's website <http://rapidintellect.com/AEQweb/fal02.htm>

Discussion

There are three parts to our discussion: we (1) relate our results to the competence and the beliefs views of EA, (2) discuss our results in terms of corrective feedback, and (3) address implications for future research.

Competence and Beliefs Views

From the competence view, high school juniors and college juniors had accurate self-efficacy; the highest performers within each group were measured as being the most highly efficacious, and the lowest performers had the lowest self-efficacy. As evinced by the correlations in Table 1, efficacy was a predictor of performance for both of the groups. However, the performances of the college freshmen varied across levels of perceived efficacy. The resulting lack of linearity between efficacy and performance in the college freshman group is inconsistent with the competence view of EA.

The beliefs view leads to an alternative interpretation of the data. Although self-efficacy was predictive of performance for the 11th grade group, many of the students statistically overestimated their performances--inaccurately high self-efficacy from a criterion-referenced perspective (mean = 37.5, or 68% correct, yet highly confident). From the beliefs view, some of the college freshmen had accurate self-efficacies and others did not. In other words, some students with high self-efficacy performed well, while other highly confident students performed poorly despite high self-efficacy (mean = 43.3, or 78%). In contrast, the self-efficacy of the college juniors was accurate from the beliefs view (mean = 47.3, or 86%). Statistical analyses revealed no differences between efficacy and performance within the group; students with the highest self-efficacy applied almost all of the punctuation marks correctly, and students who performed lower had corresponding lower levels of self-efficacy. Most college juniors had accurate perceptions of what they were capable of accomplishing.

Instructional Implications

The malleable nature of self-efficacy has been attributed, in part, to verbal persuasion (Bandura, 1997). In terms of classroom practice, verbal persuasion is perhaps manifest in the types of feedback teachers give their students. Lack of Correctional Feedback: high school juniors. Specific, correctional feedback is critical for students to form realistic perceptions of their competences (Bandura, 1997). Perhaps one factor in the inaccuracy of the self-efficacy of the high school juniors is a lack of appropriate, correctional feedback. In the school from which our 11th grade participants were pooled, it is a common practice for English teachers to abstain from penalizing students when they use improper grammar or misuse punctuation. Instead, the students are praised for engaging in the process of writing and for the written product in general terms of task completion. The desired outcome is that students are more willing to engage in writing exercises, but such praise could incorrectly influence a student's self-efficacy. General praise for work based on the quantity, or the duration of task engagement could lead a student to infer that the product was error-free when, in fact, errors exist (Bandura, 1997). Attributing a student's successful performance to effort should increase self-efficacy. However, when effort is, in fact, the only performance guideline, the increased efficacy may only motivate the student to engage and not to improve. It may foster a student's false sense of success, increase confidence for the general skill of writing, and, inadvertently, increase self-efficacy for all of its sub-skill components, including punctuation.

Presence of Correctional Feedback: college freshmen. Correctional feedback is perhaps key to the considerable variability of EA within the college freshmen group as well. The college freshmen in our study were nearing the end of their first semester in college, and each student was enrolled in a freshman-writing course. In the Meier study (1984), the completion of a similar course changed the statistical relationship between students' self-efficacy and performance. Though we found similar results, some would argue that such a change is perhaps due to a combination of factors, including the difficulty or ease with which each student met with success in other classes that same semester, their first in college. But one of the basic premises of self-efficacy is that it is domain specific and that one may have varying efficacies for different tasks, each efficacy being governed by independent experiences with each task. If one subscribes to that philosophy, one must also recognize that a course addressing a specific task would be most responsible for any change in efficacy for that task.

There are, perhaps, two possible reasons for variability in college freshmen efficacy: (1) if students had not received proper corrective feedback in high school, and then received proper corrective feedback in their first semester of college, or (2) if students were simply expected to meet higher standards of writing mastery in college. Both would lead to more scrutinizing, corrective feedback that could affect self-efficacy in two ways: (1) by providing sound remediation and (2) by altering student confidence. The combination of each could explain the variability we found in EA among the individuals in the group. If feedback is indeed corrective in nature, a student's performance can be positively affected, or can show no change. Similarly, corrective feedback can affect self-efficacy. However, efficacy can increase, decrease, or remain the same (Bandura, 1997). When combined, these effects are perhaps manifest in five possible student outcomes: (1) Lower confidence/ No learning, (2) No change in confidence/No learning, (3) Higher confidence/No learning, (4) Learning/No change in confidence, and (5) Learning/higher confidence. The presence of some or all of these effects within the college freshman group could result in the weak correlation between efficacy and performance for the college freshmen.

Future Research

During debriefing of the high school junior group, a conversation ensued between the researchers and the students. The students were curious about what we were measuring and wanted to know the answers to the punctuation task. One of the researchers explained our intentions and asked them how they thought they did; that is, how accurate they believed their perceptions would pan out to be. A discussion of the nature of the class (a class for talented and gifted students) led students to assume that they had done well. Typical quips were: "This is a gifted class, we learned that stuff in elementary school", and "If you don't know how to punctuate you shouldn't been in a gifted class." Perhaps the inaccurate nature of the high school juniors' perceptions is the result of an interaction of entity beliefs about ability and/or self-concept. An inability to separate their confidence for a specific task from their perceived academic ability (presumably versus their peers at large) might have resulted in a distortion of EA.

Also, although correlational research has suggested that causal attributions of intelligence become more accurate between elementary school and high school (Shell et al, 1995), and self-efficacy among gifted students is more accurate than others (Ewers & Wood, 1993), our findings suggest that differences may exist. Further research addressing the development of EA in these populations is needed. Finally, more investigation is needed regarding the effects of praise for task engagement on self-efficacy. Attributing success to effort has been shown to increase self-efficacy, but the standard to which high performances are judged could impact motivation. As postulated of our high school freshmen, students who are motivated to engage may not be motivated to improve if improvement is never an expected result of effort.

Conclusion

It is critical to note that, statistically speaking, there were no differences in self-efficacy among the groups--only differences in performance. A clear cause of performance differences among the groups is that each group represents a different level of education. In fact, a brief view of our results might paint an encouraging picture of the progressive nature of the participants' educations because we want students to perform better on academic tasks every year. However, recall that most students in our study were highly overconfident. Though social cognitive theory postulates that some overestimation of self-efficacy leads to increased motivation in students (Bandura, 1997), no attempt has been made to determine where the enhancement of overestimated self-efficacy ends and detriments might begin. Certainly there exists a point where exaggerated self-efficacy moves from the motivational to the delusional.

References

Bandura, A. (1997). Self-efficacy: the exercise of control. New York: Freeman.

Ewers, C.A. & Wood, N.L. (1993). Sex and ability differences in children's math and self-efficacy and prediction accuracy. Learning and Individual Differences, 5, 259-267.

Glenberg, A & Epstein, W., (1985) Calibration of Comprehension, Journal of Experimental Psychology, 11, 702-718.

Kaley, R. & Cloutier, R. (1984). Developmental determinants of self-efficacy predictiveness. Cognitive Therapy & Research, 8, 643-656.

Lachman, M.E. & Jelalian, E. (1984). Self-efficacy and attributions for intellectual performance in young and elderly adults. Journal of Gerontology, 39, 577-582.

Meier, S.T., McCarthy, P.R., Schmeck, R.R. (1984). Validity of self-efficacy as a predictor of writing performance. Cognitive Therapy & Research, 8, 107-120.

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

Olson, G. (1999). JAC: punctuation made simple. Peer reviewed, on-line Journal (Georgia State University) www.cas.edu/JAC/pms

Pajares, F. (1996). Self-efficacy beliefs in academic settings. Review of Educational Research, 66, 543-578.

Pajares, F., Miller. M., Johsson, M. (1999). Gender differences in writing self-beliefs of elementary school students. Journal of Educational Psychology, 91(1), 50-61.

Pintrich, P. & Schunk, D. (1996). Motivation in education: Theory research and applications. Upper Saddle River, N.J.: Prentice-Hall.

Pressley, M., Levin, J.R., Ghatala, E.S., and Ahmad, M. (1987). Test monitoring in young grade school children. Journal of Experimental Child Psychology, 43, 96-111.

Ryan, R. & Deci, E. (1998). Intrinsic and Extrinsic motivations: Classic definitions and new directions. Paper presented at the annual meeting of the American Educational Research Association, San Diego.

Schunk, D. H. (1989). Social Cognitive theory and self-regulated learning. In B. J. Zimmermand and D. H. Schunk (Eds.) Self-regulated learning and academic achievement: Theory, research, and practice. New York: Springer-Verlag.

Schunk, D. H. (1991). Self-efficacy and Academic motivation. Educational Psychologist, 26, 207-231.

Schunk, D. (1994). Goal and self-evaluative influences during children's mathematical skill acquisition. Paper presented at the annual meeting of the American Educational Research Association, New Orleans.

Schunk, D. & Zimmerman, B. (1998) Self-Regulated Learning. From teaching to self-reflective practice. New York: The Guilford Press.

Shell, D. F., Murphy, C. C., & Bruning, R. H. (1995). Self-efficacy, attributions, and outcome expectancy mechanism in reading and writing achievement: Grade-level and achievement-level differences. Journal of Educational Psychology, 87, 386-398.

Stone, D. (1994). Overconfidence in initial self-efficacy judgments: Effects on decision processes and performance. Organizational Behavior and Human Decision Processes, 59, 4 52-474.

Weaver, C. (1990) Constraining factors in calibration of comprehension, Journal of Experimental Psychology, 16, 214-222.

L. Brent Igo, University of Nebraska-Lincoln Michael D. Toland, University of Nebraska-Lincoln Terri Flowerday, University of New Mexico Samuel Y. Song, University of Nebraska-Lincoln Kenneth A. Kiewra, University of Nebraska-Lincoln

Dr. Flowerday was awarded the Doctor of Philosophy Degree from the University of Nebraska. Her research interests include the use of instructional choice, motivation, and learning. This study is the product of a doctoral seminar taught by Terri. Dr. Kiewra was awarded the Doctor of Philosophy Degree from Florida State University; he is currently a professor of Educational Psychology. His interests include note taking, learning strategies, and expertise. The remaining authors are graduate students.