Self-confidence and performance: A little self-doubt helps

Objectives: To test the hypothesis that a decrease in confidence on a well-learned task will increase effort and performance. Design: A 2 (group: control, experimental) 2 (trial: practice, competition) mixed-model with repeated measures on the second factor. Method: Expert skippers’ (n 1⁄4 28) self-confidence was reduced via a combination of task (i.e., change of rope) and competitive demands. Performance was the number of skips in a 1-min period. On-task effort was measured via the verbal reaction time to an auditory probe. Results: The group trial interaction (F (1, 26) 1⁄4 6.73, p < .05, h2 1⁄4 .21) supported the hypothesis: Posthoc tests revealed a significant decrease in self-confidence and a significant improvement in performance from practice to competition for the experimental group only. No significant effort effects were revealed. Conclusions: Some self-doubt can benefit performance, which calls into question the widely accepted positive linear relationship between self-confidence and performance. As effort did not increase with decreased confidence, the precise mechanisms via which self-confidence will lead to an increase or a decrease in performance remain to be elucidated. 2010 Elsevier Ltd. All rights reserved. The relationship between stress, anxiety, self-confidence and performance continues to attract much research attention (e.g., Beilock & Gray, 2007; Woodman & Hardy, 2001). Among the least disputed of these relationships is the positive association between self-confidence and performance. The support for this positive relationship is strong both theoretically and empirically (e.g., Bandura, 1997; Bandura & Locke, 2003; Martens, Vealey, & Burton, 1990; Vealey,1986, 2001;Woodman &Hardy, 2003). Meta-analyses of the self-confidence e performance relationship show that the mean effect size is greater than that revealed for cognitive anxiety and the vast majority of studies report a positive relationship between self-confidence and performance (e.g., 89% of the exact effect sizes reported in Woodman & Hardy, 2003 were positive). Bandura’s (1997) theory of self-efficacy, which is rooted in social cognitive theory, predicts a positive relationship between selfefficacy and performance by drawing on four key sources of selfefficacy that are thought to impact performance via thoughts and behaviors. Furthermore, in their development of multidimensional anxiety theory, Martens et al. (1990) theorized a positive linear relationship between self-confidence and performance. Similarly, Vealey’s (1986, 2001) sport confidence model posits a positive relationship between confidence and performance. Although the majority of research has found support for this hypothesized positive association, there exist some notable exceptions. In their study of pistol shooters, Gould, Petlichkoff, Simons, and Vevera (1987) revealed a negative relationship between selfconfidence and shooting performance. Similarly, Hardy, Woodman, and Carrington (2004) found that high self-confidence was associated with depressed golf performance scores (see also Woodman & Hardy, 2005). One explanation for such findings is that high confidence can lead to risk-taking (Campbell, Goodie, & Foster, 2004) and/or complacency (Jones, Swain, & Hardy, 1993), which in turn may hinder performance. Other models, such as Hardy’s (1996) butterfly catastrophe model, also suggest that the relationship between self-confidence and performance is not as simplistic as is commonly accepted. A further line of research utilizing awithin-person approach has revealed negative self-confidence effects. For example, in an analytical task, Vancouver, Thompson, and Williams (2001) found that over time high self-efficacy led participants to commit too * Correspondence to: Tim Woodman, Institute for the Psychology of Elite Performance, Bangor University, Gwynedd LL57 2DG, UK. Tel.: þ44 (0) 1248 382756; fax: þ44 (0) 1248 371053. E-mail address: [email protected] (T. Woodman).