An extension of the inverted-U curve hypothesis was tested by defining low, moderate, and high arousal levels as an athlete's lowest, median, and highest pregame state anxiety values across three games of a basketball tournament. Basketball performance was measured by a game statistics composite (PERF) and by total points (TP) in each game. Subjects were 30 female university varsity basketball starters from six teams. They were trichotomized on competitive trait anxiety (A-trait), and a 3 × 3 ANOVA with repeated measures on A-state categories was employed. Significant A-state effects (p < .01) were found for both dependent variables: composite game performance (PERF) and total points (TP). Although A-trait predicted absolute A-state levels extremely well (p < .001), it failed to achieve a significant relationship with performance. When intrasubject T-scores for PERF and TP were regressed separately on intrasubject A-state T-scores, the relationship of variables was seen to consist essentially of a quadratic function which explained 18.4% and 16.9% of within-subject variance for PERF and TP, respectively. High A-state scores were associated with poorest performances in all three trait groups, but plotting performance T-scores across A-state categories indicated this effect to be particularly pronounced in high competitive trait-anxiety subjects.
Special thanks are extended to Arlene C. Gorton, Associate Director of Athletics at Brown University, and to Gail Klock, former Brown Women's Varsity Basketball Coach, for their cooperation and assistance in the collection of data. Requests for reprints should be sent to Robert J. Sonstroem, Department of Physical Education, Health and Recreation, University of Rhode Island, Kingston, RI 02881.