The relative impact of cognitive anxiety and self-confidence upon sport performance: a meta-analysis TIM WOODMAN* and LEW HARDY School of Sport, Health and Exercise Sciences, University of Wales, Bangor, Gwynedd LL57 2DG, UK Accepted 21 February 2003 This meta-analysis (k = 48) investigated two relationships in competitive sport: (1) state cognitive anxiety with performance and (2) state self-confidence with performance. The cognitive anxiety mean effect size was r= 7 0.10 (P 50.05). The self-confidence mean effect size was r= 0.24 (P 50.001). A paired-samples t-test revealed that the magnitude of the self-confidence mean effect size was significantly greater than that of the cognitive anxiety mean effect size. The moderator variables for the cognitive anxiety–performance relationship were sex and standard of competition. The mean effect size for men (r = 7 0.22) was significantly greater than the mean effect size for women (r = 7 0.03). The mean effect size for high-standard competition (r = 7 0.27) was significantly greater than that for comparatively low-standard competition (r = 7 0.06). The significant moderator variables for the self-confidence–performance relationship were sex, standard of competition and measurement. The mean effect size for men (r = 0.29) was significantly greater than that for women (r = 0.04) and the mean effect size for high-standard competition (r = 0.33) was significantly greater than that for low- standard competition (r = 0.16). The mean effect size derived from studies employing the Competitive State Anxiety Inventory-2 (r = 0.19) was significantly smaller than the mean effect size derived from studies using other measures of self-confidence (r = 0.38). Measurement issues are discussed and future research directions are offered in light of the results. Keywords: cognitive anxiety, meta-analysis, self-confidence, sport performance. Introduction The relationship between anxiety and sport perfor- mance has attracted much research attention over the past 20 years, and researchers have tried to clarify this relationship by advancing several models and theories. These include multidimensional anxiety theory (Mar- tens et al., 1990), catastrophe models (Hardy, 1990, 1996a), reversal theory (Apter, 1982; Kerr, 1990) and zones of optimal functioning models (Hanin, 1980, 1986). In multidimensional anxiety theory, Martens et al. (1990) proposed a series of two-dimensional relation- ships between cognitive anxiety, somatic anxiety, self- confidence and performance. Cognitive anxiety was defined as ‘negative expectations and cognitive con- cerns about oneself, the situation at hand, and potential consequences’ (Morris et al., 1981, p. 541). Somatic anxiety was conceptualized as the perception of one’s physiological arousal. Self-confidence was conceptua- lized as one’s belief in meeting the challenge of the task to be performed. In multidimensional anxiety theory (Martens et al., 1990), cognitive anxiety is hypothesized to have a negative linear relationship with performance; somatic anxiety is hypothesized to have a quadratic (inverted-U shaped) relationship with performance; and self-confidence is hypothesized to have a positive linear relationship with performance. The hypothesized negative linear relationship be- tween cognitive anxiety and performance was largely based upon theories of attention (e.g. Wine, 1971, 1980), whereby cognitive resources are taken up by worrying thoughts and so are not available for use on the task at hand. As Martens et al. (1990) conceptua- lized cognitive anxiety and self-confidence as lying at opposite ends of a continuum, they hypothesized that self-confidence and performance would be related in a positive linear fashion. However, the rationale for the hypothesized inverted-U relationship between somatic anxiety and performance is much less clear. Martens et al. (1990) cited Weinberg’s (1978) research, which suggests that too much muscular tension will lead to a deterioration in performance. However, Martens et al. * Author to whom all correspondence should be addressed. e-mail: t.woodman@bangor.ac.uk Journal of Sports Sciences, 2003, 21, 443–457 Journal of Sports Sciences ISSN 0264-0414 print/ISSN 1466-447X online # 2003 Taylor & Francis Ltd DOI: 10.1080/0264041031000101809