A Meta-Analytic Investigation of the Within-Person Self-Efficacy Domain: Is Self-Efficacy a Product of Past Performance or a Driver of Future Performance?
Post on 09-Feb-2017
Embed Size (px)
PERSONNEL PSYCHOLOGY2013, 66, 531568
A META-ANALYTIC INVESTIGATION OF THEWITHIN-PERSON SELF-EFFICACY DOMAIN:IS SELF-EFFICACY A PRODUCT OF PASTPERFORMANCE OR A DRIVER OF FUTUREPERFORMANCE?
TRACI SITZMANNUniversity of Colorado Denver
GILLIAN YEOUniversity of Western Australia
We conducted a meta-analysis to determine whether the within-personself-efficacy/performance relationship is positive, negative, or nulland to compare the strength of the self-efficacy/performance andpast performance/self-efficacy within-person relationships. The self-efficacy/performance within-person corrected correlation was .23 butwas weak and nonsignificant ( = .06) when controlling for the lin-ear trajectory, revealing that the main effect was spurious. The pastperformance/self-efficacy within-person corrected correlation was .40and remained positive and significant ( = .30) when controlling forthe linear trajectory. The moderator results revealed that at the within-person level of analysis: (a) self-efficacy had at best a moderate, positiveeffect on performance and a null effect under other moderating condi-tions ( ranged from .02 to .33); (b) the main effect of past performanceon self-efficacy was stronger than the effect of self-efficacy on perfor-mance, even in the moderating conditions that produced the strongestself-efficacy/performance relationship; (c) the effect of past performanceon self-efficacy ranged from moderate to strong across moderating con-ditions and was statistically significant across performance tasks, con-textual factors, and methodological moderators ( ranged from .18 to.52). Overall, this suggests that self-efficacy is primarily a product of pastperformance rather than the driving force affecting future performance.
Since its inception 35 years ago, self-efficacy has become the mostfrequently studied construct in the self-regulation domain (Vancouver &Day, 2005). Self-efficacy is defined as peoples beliefs regarding theircapability to succeed and attain a given level of performance (Bandura,
We sincerely thank the researchers that contributed data to this meta-analysis. This workwas supported by Australian Research Council grants DP0984782 (Chief InvestigatorsGillian Yeo and Shayne Loft) and DP120100852 (Chief Investigators Andrew Neal, GillianYeo, and Hannes Zacher).
Correspondence and requests for reprints should be addressed to Traci Sitzmann, Univer-sity of Colorado Denver, 3920 Perry St., Denver, CO 80212; Traci.Sitzmann@ucdenver.edu.
C 2013 Wiley Periodicals, Inc. doi: 10.1111/peps.12035
532 PERSONNEL PSYCHOLOGY
1977).This construct was derived from self-efficacy theory, which pro-poses that self-efficacy enhances performance via increasing the diffi-culty of self-set goals, escalating the level of effort that is expended,and strengthening persistence (Bandura, 1977, 2012; Bandura & Locke,2003). Providing support for this notion, the overwhelming majority ofresearch has found positive relationships between self-efficacy and per-formance. Self-efficacy has been shown to increase performance by 28%,which is a stronger effect than goal setting, feedback interventions, or be-havior modifications (Stajkovic & Luthans, 1998). In addition, more than93% of studies have found positive correlations between self-efficacy andperformance at the between-persons level of analysis (Sitzmann & Ely,2011; Stajkovic & Lee, 2001).
Yet, the last decade of research has called into question two core con-clusions regarding the legacy that self-efficacy has a strong, positive effecton performance. The first question relates to the direction of this effect. Incontrast to self-efficacy theory, control theory suggests that self-efficacyseffect on performance could be positive, negative, or null depending onthe way in which self-efficacy beliefs exert their effects (Powers, 1991).The second question relates to the direction of causality and whether self-efficacy is a driver of future performance or a product of past performance.Empirical evidence suggests that the positive relationship observed at thebetween-persons level of analysis is driven by the effect of past perfor-mance on self-efficacy rather than vice versa (Beattie, Lief, Adamoulas,& Oliver, 2011; Vancouver, Thompson, & Williams, 2001). Thus, the al-most unwavering belief that self-efficacy enhances performance may bein jeopardy.
We seek to answer these questions by conducting the first meta-analysis of the self-efficacy domain at the within-person level of analysis.Control theorys core arguments relate to how confidence and performanceevolve over time, which requires an examination of dynamic, within-person processes (Powers, 1991; Vancouver et al., 2001). Further, althoughbetween-persons research is valuable for determining whether people whohave high self-efficacy outperform those with low self-efficacy, it can-not address the direction of causality between reciprocally related con-structs. Thus, we obtained data from 38 within-person self-efficacy studiesto meta-analyze the within-person relationships among self-efficacy andperformance as well as past performance and self-efficacy to assess thedirection and magnitude of these effects.
Another unique contribution of this meta-analysis is that we investi-gate the role of covariates in contributing to varied conclusions acrossprimary studies. In the self-efficacy domain, there has been little dis-cussion regarding what covariates to use, when to use them, and why;yet, it is impossible to compare results across studies without running
SITZMANN AND YEO 533
comparable analyses. Moreover, both self-efficacy and control theoriesacknowledge that the self-efficacy/performance relationship is affectedby contextual factors (Bandura, 1997, 2012; Vancouver, 2005, 2012), andbetween-persons designs preclude examining whether moderators affectthe self-efficacy/performance or past performance/self-efficacy relation-ships. We draw on arguments from this literature to examine the impactof five moderators representing the performance task, contextual factors,and methodological variables.
In the following sections, we review key propositions from self-efficacy and control theories regarding the relationships between self-efficacy and performance as well as past performance and self-efficacy.We then discuss the effects of covariates and moderators on theserelationships.
Self-Efficacy and Performance
Is the Self-Efficacy/Performance Relationship Positive, Negative, or Null?
Self-efficacy and control theories both acknowledge that discrepancycreation (positive feedback loops) and discrepancy reduction (negativefeedback loops) play a role in motivation (Bandura, 1991; Phillips,Hollenbeck, & Ilgen, 1996; Scherbaum & Vancouver, 2010), althoughthe emphasis placed on these two processes differs across theories. Dis-crepancy creation involves setting goals that are higher than peoples pre-vious best performance, whereas discrepancy reduction involves strivingto eliminate goal-performance discrepancies (Phillips et al., 1996).
Self-efficacy theory focuses on continuous improvement through dis-crepancy creation (Bandura, 1977, 1991, 1997). People with high self-efficacy are presumed to set higher goals and outperform those with lowself-efficacy. Discrepancies are created via goal setting, but discrepancyreduction is also required as one exerts effort to achieve goal mastery(Bandura & Locke, 2003). Consistent with its focus on discrepancy cre-ation, self-efficacy theory generally predicts a strong, positive effect ofself-efficacy on performance.
Control theory emphasizes the role of discrepancy reduction in regu-lating goal progress (Carver & Scheier, 1981, 1990, 2000; Powers, 1978;Vancouver, 2005). The optimistic belief that behavior is effective (i.e.,high self-efficacy) results in the current state being perceived as closer tothe goal and less effort being exerted toward goal accomplishment thanwhen self-efficacy is low (Powers, 1973; Vancouver & Kendall, 2011).This process could produce a null or negative effect of self-efficacy onperformance, such as when self-efficacy beliefs are inflated relative to
534 PERSONNEL PSYCHOLOGY
actual performance levels or performance levels are ambiguous.1 Controltheory also acknowledges that goals operate within a hierarchical system,such that people engage in discrepancy creation for lower level goals toattain higher level goals (Carver & Scheier, 1981, 2000; Phillips et al.,1996; Scherbaum & Vancouver, 2010).
To understand the conditions under which self-efficacy produces pos-itive, negative, or null effects on performance, the relationship should beexamined at the within-person level of analysis (Powers, 1991; Vancouveret al., 2001). Theoretically, self-regulation is a within-person process thatevolves over time as people establish goals, assess their confidence forachieving their goals, exert effort, and subsequently modify their regu-latory processes (Carver & Scheier, 2000; Kanfer & Ackerman, 1989;Sitzmann & Ely, 2011). Furthermore, control theorys explanation for thevaried effects of self-efficacy are produced by dynamic, within-personprocessesthey concern how fluctuations in self-efficacy within an in-dividual relate to fluctuations in performance (Powers, 1991; Vancouveret al., 2001).
In 2001, Vancouver and colleagues demonstrated a negative effect ofself-efficacy on performance in a context in which the constructs fluctuatedover time within-individuals and performance levels were ambiguous.This sparked a flurry of studies examining the within-person relationshipbetween self-efficacy and performance (e.g., Feltz, Chow, & Hepler, 2008;Richard, Diefendorff, & Martin, 2006; Schmidt & DeShon, 2010; Seo &Ilies, 2009; Vancouver, Thompson, Tischner, & Putka, 2002; Vancouver& Kendall, 2011; Yeo & Neal, 2006). Consistent with the notion thatthe self-efficacy/performance relationship may not be uniformly positive,these studies demonstrated a mix of positive, negative, and null effects.
What Is the Direction of Causality Between Self-Efficacy and Performance?
Consideration of within-person self-efficacy/performance relation-ships raises the issue of reciprocal effectspast performance can affectself-efficacy, which, in turn, can affect subsequent performance. Self-efficacy and control theories agree that past performance has a positive ef-fect on self-efficacy. Past performance is positively related to self-efficacybecause information regarding how one performed in the past can be usedto judge ones capacity to succeed in the future (Ackerman, Kanfer, &Goff, 1995; Bandura, 1997; Kozlowski et al., 2001; Mitchell, Hopper,
1Note that this negative effect is expected to occur within a goal level; that is, duringthe goal striving phase when discrepancy reduction processes are active. When goal settingprocesses are active, self-efficacy is expected to have a positive effect on goal level andperformance as a function of discrepancy creation (Vancouver, More, & Yoder, 2008).
SITZMANN AND YEO 535
Daniels, George-Falvy, & James, 1994). When people initially approacha complex, novel task, they reflect on their past performance in similarsituations when judging their self-efficacy (Bandura, 1991, 1997; Wood &Bandura, 1989). In addition, feedback on past performance is comparedto ones goal level to assess goalperformance discrepancies (Carver &Scheier, 2000). High performance suggests that goalperformance dis-crepancies are small, increasing confidence in goal attainment.
Most research on the self-efficacy/performance relationship has beenconducted at the between-persons level of analysis, and between-personsdesigns cannot distinguish between the effects of self-efficacy on per-formance versus past performance on self-efficacy. As such, the strong,positive, between-persons self-efficacy/performance relationship may bea product of the positive effect of past performance on self-efficacy(Vancouver et al., 2001). Within-person designs are essential for dis-entangling the relative magnitude of reciprocal relationships.
Four within-person studies have examined these reciprocal relation-ships. Vancouver and colleagues (2001) conducted two studies and foundpast performance had a strong, positive effect on self-efficacy, whereasself-efficacy had a weak, negative effect on performance. Similarly, acrosstwo within-person studies, Beattie and colleagues (2011) found past per-formance had a strong, positive effect on self-efficacy and accounted forup to 49% of the variance in self-efficacy. Self-efficacy, in turn, had aweak, negative effect on performance and explained up to 3% of thevariance in performance. In combination, the evidence presented in thissection supports the notion that self-efficacy may primarily be a productof past performance rather than a mobilizer of future performance.
Relative Magnitude of Within- and Between-Persons Effects
Since 2001, researchers have examined the effect of self-efficacy onperformance and vice versa (albeit to a lesser extent) at the within-personlevel of analysis across a variety of performance domains.2 Table 1 reportsthe studies included in the meta-analysis, study information, and bothbetween- and within-person correlations. We do not propose a main effecthypothesis for the direction of the self-efficacy/performance within-personrelationship due to contrasting theoretical arguments and mixed empiricalresults. However, the between-persons and past performance/self-efficacy
2Although some studies conducted prior to 2001 collected repeated measurements ofself-efficacy and performance (e.g., Bandura & Jourden, 1991; Bandura & Wood, 1989), therelationship between self-efficacy and performance was examined at the between-personslevel of analysis. These studies were not included in the meta-analysis because we wereunable to obtain access to the data or within-person correlation matrices.
536 PERSONNEL PSYCHOLOGYTA