Finding a Between-Person Negative Effect of Self-Efficacy on Performance: Not Just a Within-Person Effect Anymore

Abstract

Research examining the relationship between self-efficacy and performance has tended to find a positive effect at the between-person level and a weak negative or null effect at the within-person level. This pattern of results has led to a belief that the level of analysis is a key factor in determining the sign of the relationship. Using a between-person experimental design, the authors demonstrate a negative effect for self-efficacy on motivation and quality of performance at the between-person level. A positive effect for self-efficacy on the quantity of tasks engaged, apparently due to the reduction in motivation on each task, was also found. The discussion focused on the relevance of the findings in the debate between social cognitive and control theorists.

Notes

1. ¹We should note that Vancouver et al. (2008) included a between-person manipulation that also found a negative self-efficacy effect on effort allocated to accepted goals. However, the manipulation was based on the order of task difficulty presented during a training phase and the negative effect was very small (i.e., 0.5% variance explained). This was likely due to the weakness of the manipulation in the first place (i.e., accounting for 3.6% of variance in the self-efficacy measure given immediately after the practice trials) as well as the calibrating effect of subsequent vertical, trial-level feedback during the many experimental trials.

2. ²Other conceptualizations of effort (e.g., attention/concentration; physical effort) might have the same conservation effect depending on their nature. For example, attention is more likely to have a conservation effect in a dual task scenario where both tasks are sensitive to the attention resource (e.g., Vancouver & Tischner, 2004). The effect of effort/resource conservation on sequential task performance for attention and physical effort would likely depend on the effect of fatigue on performance and the time it takes the resource to recover.

3. ³Bandura’s (2012) more recent writings appear to take a different position on this point (see Vancouver, 2012).

4. ⁴Goal-choice processes are also involved in disengaging from an accepted goal. That is, individuals might stop working on the task because they give up as opposed to thinking they have reached the goal. However, the ambiguous feedback should reduce the chance that individuals would give up because they do not really know where they stand. Stopping because one is giving up is more likely in protocols where feedback is present, and it indicates little or no goal progress.

5. ⁵One might notice that the correlation between the self-efficacy manipulation check and performance quality, though negative, was not significant. This is probably because the correlations between self-efficacy and performance quality within each experimental group averaged .35, which counteracted the between-group differences. The positive, within-group correlation likely reflects the typical confounding of actual capacity with belief in capacity. That is, better anagram solvers found a higher percentage of anagrams within their respective conditions, creating systematic covariance, as opposed to random noise, that obscures the causal processes.

6. ⁶We thank an anonymous reviewer for bringing this issue to our attention.