Scoring Bias in Springboard and Platform Diving: A Bayesian Approach to Revision of DD Tables

Abstract

Springboard and platform diving, unlike other sports competitions that use absolute performance measures, relies instead on a panel of judges to assess the aesthetic and technical quality of the athlete's performance. Using data from the 1983 U.S.A. Indoor Diving Nationals, regression analysis was used to determine if there was a statistically significant relationship between degree of difficulty (DD) and scores awarded for a dive. Surprisingly, the analysis revealed that as divers attempted higher degree of difficulty dives, judges became less inclined to award lower scores. Even though the average scores were lower for high DD dives, they were not low enough to compensate for the higher DD rating. Further analysis revealed that the women's competition had less scoring bias towards high DD dives than the men's competition. Even a small amount of scoring bias towards high DD dives is important since typically small total point differences in scores separate the finishing order of divers at a national or international competition. Bayesian analysis, based upon actual diver performance at national and international competitions, is suggested as one method of dealing with the problem of scoring bias towards high DD dives. Using this approach, probability of a score given the DD [P(DD/Score)] is derived from the initially observed P(Scores/DD). Monte Carlo methods can then be used to simulate large numbers of divers performing at each DD level to obtain more precise DD measures for a particular dive.     

Predicting Physical Activity Among Children: Investigating Interaction Effects in the Theory of Planned Behavior

ABSTRACT

Purpose: The theory of planned behavior (TPB) has been criticized for not including interactions between the variables assumed to predict behavior. This study sought to test how TPB variables interact to predict physical activity (PA) in children. Method: Four hundred thirty-eight children (M_age = 8.6 years, SD = 1.6 years) completed a TPB questionnaire and a PA questionnaire at Time 1. The PA measure was repeated 2 months later. Path analyses were performed to test the hypothesized model including interaction terms between TPB variables. Simple slopes analyses were also carried out to examine the statistically significant interaction terms. Results: Path analyses confirmed the classical hypotheses of TPB (R² for intentions = .39, R² for PA = .12) and also demonstrated only statistically significant Attitudes × Perceived Behavioral Control and Subjective Norms × Attitudes interactions (R² change for intentions = .01, p = .009). Simple slopes analyses revealed that the strength of the association between perceived behavioral control and intentions was only statistically significantly higher (t = 2.18, p = .05, d = 0.34, 95% CI [0.03, 0.65]) when attitudes were high compared with when attitudes were low. The link between attitudes and intentions was only statistically significant at a low level of subjective norms but not at a high level. Conclusion: The integration of interaction effects between TPB variables did not increase for the variance of PA explained by the model. More research appears to be necessary to explore how the TPB could be augmented to better predict PA in children.