Formative vs Reflective constructs: a CTA-PLS approach on a goalkeepers’ performance model

Nowadays, PLS-SEM is a trend-topic, whereas football is moving towards a data-driven approach; by combining these two worlds, we aim to show a new way for measuring football goalkeepers’ performance, by using data provided from EA Sports experts and available on the Kaggle data science platform. Furthermore, another objective is to refine the model, supporting football experts from a statistical point of view. For this purpose, we adopt a confirmatory tetrad analysis (CTA-PLS) to validate and evaluate the nature (e.g. formative or reflective) of each latent variable. Then, a second-order PLS-SEM model is built. We validate and compare this new indicator with a benchmark (the EA overall). The final goal is to prove the CTA approach on a real case study and to refine a composite performance indicator for helping football policy makers taking strategic decisions.