6533b7dcfe1ef96bd1272c17

RESEARCH PRODUCT

Making Every "Point" Count: Identifying the Key Determinants of Team Success in Elite Men’s Wheelchair Basketball

John FrancisAlun OwenDerek M. PetersDerek. M. Peters

subject

Basketballlcsh:BF1-990Applied psychologyLogistic regression050105 experimental psychologyOddsData modelingRC120003 medical and health sciences0302 clinical medicineParalympicPsychology0501 psychology and cognitive sciencesCategorical variableGeneral PsychologyOriginal Researchlogistic regression05 social sciencesOffensiveVDP::Medisinske Fag: 700::Idrettsmedisinske fag: 850sport performance analysisEuropean championshipslcsh:PsychologyElitePsychologypredictive modeling030217 neurology & neurosurgeryPredictive modelling

description

Wheelchair basketball coaches and researchers have typically relied on box score data and the Comprehensive Basketball Grading System to inform practice, however, these data do not acknowledge how the dynamic perspectives of teams change, vary and adapt during possessions in relation to the outcome of a game. Therefore, this study aimed to identify the key dynamic variables associated with team success in elite men’s wheelchair basketball and explore the impact of each key dynamic variable upon the outcome of performance through the use of binary logistic regression modelling. The valid and reliable template developed by Francis, Owen and Peters (2019) was used to analyse video footage in SportsCode from 31 games at the men’s 2015 European Wheelchair Basketball Championships. The 31 games resulted in 6,126 rows of data which were exported and converted into a CSV file, analysed using R (R Core Team 2015) and subjected to a data modelling process. Chi-square analyses identified significant (p<0.05) relationship between Game Outcome and 19 Categorical Predictor Variables. Automated stepwise binary regression model building was completed using 70% of the data (4,282 possessions) and produced a model that included 12 Categorical Predictor Variables. The accuracy of the developed model was deemed to be acceptable at accurately predicting the remaining 30% of the data (1,844 possessions) and produced an area under the receiver operating characteristic curve value of 0.759. The model identified the odds of winning are more than double when the team in possession are in a state of winning at the start of the possession are increased five-fold when the offensive team do not use a 1.0 or 1.5 classified player but are increased six-fold when the offensive team use three or more 3.0 or 3.5 players The final model can be used by coaches, players and support staff to devise training and game strategies that involve selecting the most appropriate offensive and defensive approaches when performing ball possessions to enhance the likelihood of winning in elite men’s wheelchair basketball.

10.3389/fpsyg.2019.01431https://eprints.worc.ac.uk/8119/1/fpsyg-10-01431.pdf