Self-efficacy is defined as the levels of confidence individuals have in their ability to execute courses of action or attain specific performance outcomes (Bandura, 1977, 1982, 1997). Personal efficacy expectations are proposed to influence initiating behavior, how much effort will be applied to attain an outcome, and the level of persistence applied to the task in the face of difficulties and setbacks (Bandura, 1997). The veracity of this claim has contributed to a great deal of research effort and interest from applied practitioners. It is generally agreed that applied interventions should be founded on a basis of theory and research. The purpose in the present study was to test the self-efficacy– performance relationship in an academic setting.

Research findings are generally consistent with the notion that high self-efficacy is associated with successful performance, although the strength of relationships varies between studies. Bandura (1997) argued that for self-efficacy to predict performance outcome, self-efficacy estimates should be made toward factors important to the attainment of the behavior of interest. Self-efficacy measures and performance should lie in the same behavioral domain. It is important, for example, that the competences that researchers select as the basis for self-efficacy ratings should be the competences required in delivering subsequent performance (Lachman & Leff, 1989; Pajares, 1996; Pajares & Miller, 1995). Thus, self-efficacy research should involve a thorough examination of the competences that underpin performance.

A second factor proposed to influence the predictive power of self-efficacy is the knowledge that participants have regarding the task (Bandura, 1997). Complex tasks involving heavy demands on knowledge, cognitive ability, physical and mental effort, behavioral skills, and persistence present real difficulties for accurate self-efficacy estimates, particularly if participants have had no previous experience of such complex tasks (Multon, Brown, & Lent; 1991; Stajkovic & Luthans, 1998). In a meta-analysis of self-efficacy in an academic setting, Multon et al. found that self-efficacy and performance relationships were the strongest on tasks that were relatively short in duration and for which the measure of self-efficacy was taken shortly before performance.

Collectively, there has been a great deal of research support for the central tenets of self-efficacy theory (Bandura, 1997). Research findings show the weakest self-efficacy and performance relationships are in field studies involving complex tasks. If self-efficacy research is to impact on real-world settings that typically involve complex tasks, there is a need for well-designed research aimed at investigating self-efficacy and performance relationships in ecologically valid settings. The purpose in this study was twofold: the first purpose was to identify measures of self-efficacy that tap into the perceived capability of the full range of behaviors of interest; the second purpose was to investigate whether self-efficacy measures can predict academic performance some months after the measures of self-efficacy have been obtained. This is not to suggest that self-efficacy could be used for the purpose of selecting students on to programs as there is a high likelihood of leniency or of overstatement of efficacy expectations if students knew that entry decision rested in part on their own self-efficacy estimates. However, research showing significant self-efficacy and performance relationships could be used as an empirical base for intervention strategies designed to raise self-efficacy to bring about improved performance.

Development of a self-efficacy measure
The research strategy was to develop a self-efficacy measure to assess confidence intervals toward the competences needed to achieve success in the course. Lecturers who taught on the modules listed the competences they believed would be needed to achieve success in the module. This list of competences was given to a group of 10 students (5 full-time and 5 part-time). In group sessions, students were asked to consider what they thought would be the most critical aspects and competences needed to succeed as a student. Group session results identified that having the intellectual ability to cope with course content, and being able to manage time were the most important competences. This latter competence is perhaps particularly relevant for part-time students who need to juggle the demands of study with the demands of full-time employment. The ability to manage time is also relevant as far as full-timers are concerned, as many of them need to supplement student loans with evening or weekend employment. Successful completion of the course was perceived to depend on the ability to either make time, or use available time for study.

Participants were asked to complete a number of self-efficacy measures on a scale of 1–10, where 10 = total confidence and 1 = no confidence. Questions included: How confident are you today on the scale of 1–10 that: (1) You can cope with the intellectual demands of the program; (2) You can make sufficient effort to meet the demands of the program; (3) You can manage your time to meet the demands of the program; (4) You will pass assignments/exams first time – i.e. no re-sits; (5) You will attain average grades for semester 1 modules of 1–4 (70%+); (6) You will attain average grades for semester 1 modules of 5–7 (60-69%); (7) You will attain average grades for semester 1 modules of 8–16 (40-59%).

Reliability of self-efficacy measures
The issue of the reliability (stability) of self-efficacy ratings is a thorny one and rarely investigated. By its very nature self-efficacy is malleable, being influenced by previous successful accomplishments, vicarious experiences, verbal encouragement, and affective feelings (Bandura, 1982, 1997). Self-efficacy measures based on performance accomplishments are likely to be more stable than self-efficacy measures based on affective states. In the present study, self-efficacy measures were taken twice, separated by one week. Students received no feedback or further details of the course over that period; hence it is reasonable to assume that self-efficacy would remain stable. By contrast, efficacy measures that varied significantly might be more a reflection of mood rather than an assessment of ability to achieve desired outcomes. It is acknowledged that if the time between tests is too brief then students may well remember their previous estimates with the resultant analysis representing strength of memory trace rather than a true estimate of personal capability.

Measure of performance outcome
The measure of performance was the summation of grades obtained over two modules. Many of the other modules were not formally assessed, or involved marks being awarded on a group basis – i.e., same mark to each group member – and these modules were not included in the study. All work was marked by the class tutor, second marked, and 20% of the work was then finally marked by an external examiner. It was felt that this would provide an ecologically valid indication of performance.

Grades were on a scale of 1–16 where 1 was the highest pass, and 16 was the lowest pass, however, as the measure of performance used was the summation of 2 modules, grades could range from 2 to 32. In both modules, assessment comprised a written assignment and a written examination. The written assignment was submitted in week 13 and the examination was sat in week 14.

Descriptive statistics for self-efficacy measures over time and academic performance are contained in Table 1. As shown in Table 1 participants were relatively high in confidence toward coping with the intellectual demands of the program, in making sufficient effort to meet the demands of the program, in managing time to meet the demands of the program, in passing assignments/exams first time, and in attaining average grades for semester 1 modules of 8–16.

Participants were less confident regarding attaining average grades for semester 1 modules of 1–4 (70%) and attaining average grades for semester 1 modules of 5–7 (60-69%). The mean mark for academic performance was a grade of 13.3 (SD = 5.45), thus indicating that students achieved a pass mark towards the high (in performance) end on a scale that ranged from 2 to 32.

Reliable self-efficacy measures were entered into a standard multiple regression equation, with the end of semester performance marks as the dependant variable. Regression results showed that 11.5% (Adj. R2 = 0.115, p < .05) of the variance in performance was explained by self-efficacy measures. Results indicated that “cope with the intellectual demands of the program” was the only significant predictor (r = -0.40, p < 0.01). The direction of the relationship indicated that as self-efficacy scores tended to increase, academic performance improved. The other two measures, “pass first time” and “achieve grades between 1–4” showed no significant relationship with academic performance.

The present study had two purposes: to develop a self-efficacy measure that tapped the perceived capability of the full range of behaviors of interest, and to investigate whether the self-efficacy measure predicted performance approximately 13 weeks after the measure was administered. Interviews were used to identify the important behaviors that influence performance, hence suggesting that self-efficacy items reflected the behaviors students perceive as influencing academic performance.

Results showed that stable self-efficacy measures were associated with 11.5% of performance variance with “confidence to cope with the intellectual demands of the program” as the only significant predictor. It is important to note that this study had none of the conditions said to maximize the self-efficacy and performance relationship as there was a 13-week period between the time students completed the self-efficacy questionnaire and the performance measure. Furthermore, studying for postgraduate qualifications is proposed to represent a highly complex set of related factors. Judgments of self-efficacy toward coping with the cognitive demands of the program are especially difficult in the early stages of the program. Additionally, there are the factors of time management, particularly in the case of part-time students, who need to juggle the demands of paid employment with the demands of study, attendance at classes, family, and social pressures. Collectively, it is suggested that the self-efficacy and performance relationship (r = -0.40) compares favorably with previous research in which support for the predictive effectiveness of self-efficacy measures has been found (Multon et al. 1991; Stajkovic & Luthans, 1998).

Self-efficacy theory offers an attractive construct on which to base applied interventions due to the proposed efficacy-performance relationship. Self-efficacy is also malleable to the effects of positive and negative performance feedback: the implication being that performance can be improved via positive performance feedback that raises efficacy expectations (Podsakoff & Farh, 1989). A necessary condition for self-efficacy to be used as an effective intervention tool is the existence of significant relationships between the self-efficacy measures and the performance criteria. Moreover, the longer the time period between the self-efficacy measure and performance, the greater the scope and time for intervention. The identification of a valid self-efficacy measure at the start of an academic year would allow scope for many weeks of teaching etc. prior to performance at, for example, the end of the semester. In the present study, findings indicate that interventions to enhance self-efficacy to cope with the intellectual demands could be considered.

There are several different strategies that educational practitioners could use to enhance self-efficacy toward intellectual ability. The notion of what intellectual abilities are needed to pass the course impacts on the cognitive process of making efficacy expectations towards attaining them. Providing students with clear knowledge of the task is one strategy that could be implemented. One method of improving knowledge of the task is to provide clear assessment guidelines. A second method could be for students to be shown work that has passed the program. A third method could be for incoming students to meet students who successfully completed the course to discuss the competences needed for success.

In conclusion, in the present study we developed a measure of self-efficacy that was valid for the context of the sample under investigation. Regression results showed that self-efficacy toward intellectual ability predicted subsequent academic performance. Importantly, this relationship was found even though the time gap between self-efficacy and performance was long, and the complexity of the task was high. Both of these are factors previously found to reduce the strength of the self-efficacy and performance relationship. It is suggested that future research is needed to cross-validate the findings of the present study to a new sample.