Three goals of research | |||
To describe | To predict | To draw causal inference | |
Research objective can be | To describe how athletes perform their training or how many athletes that sustain injury over a given time span | To predict who is at risk of injury (by comparing athlete workloads and characteristics) | To understand why injury occurs (by comparing athlete workloads and characteristics) |
Researchers aim to | Describe, not to investigate associations | Investigate an association which is non-causal | Investigate a causal association |
Researchers can use | At least one variable in the analysis* | At least two variables in the analysis | At least two variables in the analysis |
Role of additional variables | No additional variables are included in the analysis | Subgroup stratification (optional) | Confounders, mediators and/or effect-measure modifiers |
Should the analysis be based on frameworks for sports injury occurrence? | No | No | Yes |
Some examples of statistical techniques | Regression Cluster analysis … | Regression Classification … | Regression IPW IV-estimation G-estimation G-formula Standardisation … |
Type of study | Exploratory or confirmatory | Exploratory or confirmatory | Exploratory or confirmatory |
Goal allows researchers | To describe a workload-related or an injury-related variable | To predict which athletes are most vulnerable to injury. These athletes may be target population for interventions. | To identify potential targets for sports injury prevention. Here, authors assess the role of causal factors which if targeted for interventions, are likely to reduce sports injury risk. |
Classification includes, but is not limited to, decision trees, random forests, neural networks.
*If the aim is to describe injury development over time, two variables are needed (injury and time). Please note that the term ‘how’, which is used in relation to describing in the table, has previously been used in the context of causal inference in relation to the mechanisms underpinning sports injury occurrence in other educational pieces.20
IPW, inverse probability weighting; IV estimation, Instrumental variable estimation.