Investigating the coach's eye when evaluating and selecting 3 × 3 basketball players

Although recent studies recommend making multidimensional selection decisions based on subjective and objective data, in practical settings coaches often decide subjectively. When trying to understand coaches' selection decisions, it is important to regard different theories of decision-making when analyzing predictors. Thus, this study examined predictors for selection decisions by comparing approaches of the theory of both unbounded (logistic regression) and bounded rationality (fast-and-frugal tree, FFTree). Accordingly, this study's aim was twofold: (1) investigate talent aspects in 3 × 3 basketball, and (2) compare different statistical approaches to this investigation. Regarding the underlying process of selection decisions, analyses focused on interrater agreement of subjective assessments and the relevance of different talent aspects. A total of 46 3 × 3 basketball players (female n = 23, age range: 15.8–18.6 years) were assessed during the final stage of the German national selection system. Data included motor performance tests, anthropometrics, and coaches' subjective assessments of technical, tactical, and psychological parameters, as well as ratings of the relevance of different talent aspects. Variables were first reduced and then analyzed using logistic regression and FFTree. Interrater reliability was assessed via intraclass correlation coefficients (ICC), and graphical analyses examined talent aspects relevance. Both logistic regression (Nagelkerke R 2 = 0.70) and FFTree indicated that the subjective assessments component was the most important predictor for selection. The ICC revealed poor to moderate consistency between coaches (female players = 0.38; male players = 0.51). Graphical comparisons and descriptive statistics suggest differences between national and regional coaches in their assessments of the relevance of different talent aspects such as current performance. Although logistic regression and FFTree yielded similar main results, FFTree visualized and identified a prioritization between cues. Therefore, FFTrees are considered as an appropriate method for future studies on selection decisions, as they reflect decision-making under limited information and time pressure present in selection settings. Player selection in 3 × 3 basketball appears highly subjective, as indicated by low interrater reliability between coaches. To enhance quality and transparency, selection decisions should be made by a team of coaches based on clearer, shared criteria and be supported by actuarial approaches such as FFTrees.