Dyadic analysis for multi-block data in sport surveys analytics
针对运动员与教练的成对数据，提出一种替代现有模型的量化回归方法，分析影响游泳运动员表现的心理因素，为教练和运动员制定策略提供依据。
Abstract Analyzing sports data has become a challenging issue as it involves not standard data structures coming from several sources and with different formats, being often high dimensional and complex. This paper deals with a dyadic structure (athletes/coaches), characterized by a large number of manifest and latent variables. Data were collected in a survey administered within a joint project of University of Naples Federico II and Italian Swimmer Federation. The survey gathers information about psychosocial aspects influencing swimmers’ performance. The paper introduces a data processing method for dyadic data by presenting an alternative approach with respect to the current used models and provides an analysis of psychological factors affecting the actor/partner interdependence by means of a quantile regression. The obtained results could be an asset to design strategies and actions both for coaches and swimmers establishing an original use of statistical methods for analysing athletes psychological behaviour.