Reporting of generalized linear mixed models (GLMM) in sports sciences: a scoping review
Statistical methods are essential in sports sciences for decision-making in performance analysis, injury prevention, and athlete outcomes. Generalized Linear Mixed Models (GLMMs) are widely used to estimate fixed and random effects, particularly when dependent variables are binary, ordinal, count, o...
| Autores: | , , , , |
|---|---|
| Tipo de recurso: | artículo |
| Fecha de publicación: | 2025 |
| País: | España |
| Institución: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/451576 |
| Acceso en línea: | https://hdl.handle.net/2117/451576 https://dx.doi.org/10.1177/22150218251384557 |
| Access Level: | acceso abierto |
| Palabra clave: | Generalized linear mixed models Scoping review Sports sciences Methodological quality Àrees temàtiques de la UPC::Matemàtiques i estadística |
| Sumario: | Statistical methods are essential in sports sciences for decision-making in performance analysis, injury prevention, and athlete outcomes. Generalized Linear Mixed Models (GLMMs) are widely used to estimate fixed and random effects, particularly when dependent variables are binary, ordinal, count, or non-normally distributed quantitative data. Alternative models, such as Vector Generalized Additive Models (VGAM) and transformation mixed-effects models (tramME), may also be appropriate for specific data structures, especially in repeated measures contexts. This scoping review, following PRISMA guidelines, examines the use and reporting of GLMMs in sports sciences. A search of articles published before March 4, 2023, identified 55 studies from databases such as PubMed and Web of Science. GLMMs were primarily applied in soccer (20%) and multidisciplinary sports (16.4%). The most common response variable distributions were Poisson and Binary (25.7% each), while overdispersion was not evaluated in 75% of studies. R was the most frequently used software (41.8%), but only 34.3% of articles specified the statistical package. Data and/or code sharing was reported in 17.1% of articles. Most important information about GLMMs was not reported in most articles, indicating a need to improve the quality of reporting in line with current recommendations for the use of GLMMs. |
|---|