Project
Predicting Performance in the Complex Soccer Environment
My PhD project is concerned with gaining insight into performance assessment and prediction in the context of soccer. I will focus on important methodological and psychometric issues that limit the understanding of the psychological tests and behavioral observations employed for the assessment of (future) sport performance. The need for resolving these issues is highlighted by the fact that few robust predictors of soccer performance have been identified despite decades of research (Johnston et al., 2018). In my PhD project, I aim to address a variety of these issues by combining insights from psychometrics, selection psychology and sport sciences. Notably, although this PhD focuses specifically on the context of soccer, it assumes that the principles derived from it are applicable to a wide-range of performance settings such as education and work.
One major statistical issue is the ignorance of the influence of range restriction (Bergkamp et al., 2019). In most previous soccer talent identification studies, participants were recruited from pre-selected and homogeneous groups of skilled players, such as youth players at professional clubs. Consequently, the predictive validity of psychological tests (e.g., self-regulation or motivation) for sports performance were typically underestimated, but to unknown and varying degrees. To draw conclusions about the validity of psychological tests at the population level, it is important to investigate them in a large heterogeneous group of players. Accordingly, I will study the predictive validity of a variety of psychological tests (e.g., self-regulation) and motor skill tests (e.g., Körperkoordinationstest für Kinder) for an in-game soccer performance criterion in a heterogeneous group of youth soccer players with data made available by the Royal Dutch Football Association (KNVB).
A second issue concerns the standard practice of only measuring distinct skills and characteristics, such as the previously mentioned self-regulation test. Measuring such isolated characteristics as predictors is called the “signs” approach in selection psychology (Wernimont & Campbell, 1968) which is the traditional concern of psychometrics. While signs could be valuable in predicting performance in heterogeneous groups of players, this approach is typically less effective when selecting from homogenous, pre-selected groups (Sackett et al., 2017). Research in other contexts like education already showed that using representative “samples” of future performance – tasks that mimic the criterion behavior – work better than using signs in homogeneous groups (Niessen et al., 2016). Accordingly, my third study will be the first to systematically compare the predictive validity of the traditional signs to the alternative samples in a homogeneous group of players. Data will be collected at a professional local soccer academy.
A third issue revolves around the reliability and validity of observational measures made by scouts. Although sports scientists rarely employed ‘samples’, scouts heavily rely on them; they often observe a player during a game to judge and predict their performance (Bergkamp et al., 2022). However, observing only one game as a sample lacks situational representativeness (Brunswik, 1952). Simply put, the situations that a player happens to encounter in one particular game, may not represent the irregular environmental conditions that a player needs to deal with across other matches and circumstances. Higher situational representativeness may be accomplished by ’sampling’ performance from a multitude of situations (e.g., against stronger, equal, and weaker opponents). Therefore, one study focuses on whether taking multiple representative samples of situations can significantly improve the observational measures made by soccer scouts.
As a whole this PhD project contributes to a better understanding of the validity of psychological tests and behavioral observations for predicting performance in the soccer context by addressing a variety of psychometrics issues. Accordingly, I believe that it is a good fit with IOPS.
Supervisors
Prof. Dr. R.R. Meijer
Dr. A.S.M. Niessen
Prof. Dr. J.R. den Hartigh
Period
1 October 2023 – 1 October 2027