Path analysis of a self-esteem model across a competitive swim season

An adaptation of the previously developed Exercise and Self-Esteem Model (Sonstroem & Morgan, 1989) was tested longitudinally with 93 male interscholastic swimmers from nine high schools who were evaluated at pre-, mid-, and postseason. Swimmers completed three self-perception scales that ranged from evaluations of specific swim skills (SKILL) through broader perceived physical competence (PC) to global self-esteem (SE). Performance scores (PERF) were calculated across events by standardizing swim times to a mean of 0 and a standard deviation of 1. Structural modeling analysis across the three time waves indicated an excellent data fit (chi 2 = 25.46, p > .15). The model was able to explain 84, 83, and 80% of PC, SKILL, and PERF, respectively, at the third testing period. Swim improvement tended to be small (median change = 2.9%). Relationships among variables at each testing period and among the same variables at different testing periods were large and as hypothesized. Several significant relationships occurred between different variables across different testing periods. These were not extinguished when social desirability was added to the model. Social desirability failed to significantly influence subsequent measures of any model variable.