4 98 CHAPTER 4 6. CONCLUSION Our study set out to provide a roadmap for causal inference and introduce the use of TMLE in sports science for other sports scientists. We applied the causal roadmap and showed that TMLE has a lower bias than GLM in a simulation setting both on the correct and the misspecified causal model. This result indicates that TMLE can be a more precise method than GLM in identifying and correctly estimating causal effects. Furthermore, when applying GLM and TMLE on the observed data on substitution, both methods found that the total physical performance improves when a substitution is made. However, the difference in the effect sizes between the correctly specified and the missspecified model was considerable for TMLE and GLM. Furthermore, we showed that in these cases, TMLE was more precise than GLM. 7. PRACTICAL IMPLICATIONS These findings show that the power of TMLE can help bring causal inference in sports science to the next level when more factors are taken into account. Future work will need to collect as much factor data as possible, enabling investigation of the influence of one factor in contrast with the traditional statistical methods where a selection of factors is made. Funding This research was partly funded by an SNN (Samenwerking Noord Nederland) MIT Grant under project code MITH20138. Institutional Review Board Statement The study was conducted according to the guidelines of the Declaration of Helsinki and approved by The Ethics Committee CTc UMCG of the University Medical Center Groningen, The Netherlands (protocol code: 201800430, 01/11/2018). 201800430 Informed Consent Statement Informed consent was obtained from all subjects involved in the study. Data Availability Statement The data can be found on Github:https://github.com/dijkhuist/EntropyTMLESubstitutions/tree/main/Data Acknowledgments The authors thank Prof. Dr. K.A.P.M. Lemmink, Prof. Dr. M. Aiello, Prof. Dr. H. Velthuijsen,
RkJQdWJsaXNoZXIy MjY0ODMw