Search results

Transfering Targeted Maximum Likelihood Estimation for Causal Inference into Sports Science

Although causal inference has shown great value in estimating effect sizes in, for instance, physics, medical studies, and economics, it is rarely used in sports science. Targeted Maximum Likelihood Estimation (TMLE) is a modern method for performing causal inference. TMLE is forgiving in the misspecification of the causal model and improves the estimation of effect sizes using machine-learning methods. We demonstrate the advantage of TMLE in sports science by comparing the calculated effect size with a Generalized Linear Model (GLM). In this study, we introduce TMLE and provide a roadmap for making causal inference and apply the roadmap along with the methods mentioned above in a simulation study and case study investigating the influence of substitutions on the physical performance of the entire soccer team (i.e., the effect size of substitutions on the total physical performance). We construct a causal model, a misspecified causal model, a simulation dataset, and an observed tracking dataset of individual players from 302 elite soccer matches. The simulation dataset results show that TMLE outperforms GLM in estimating the effect size of the substitutions on the total physical performance. Furthermore, TMLE is most robust against model misspecification in both the simulation and the tracking dataset. However, independent of the method used in the tracking dataset, it was found that substitutes increase the physical performance of the entire soccer team.

Multi-period research designs for identifying causal effects of built environment characteristics on travel behaviour

To date, most empirical studies have applied cross-sectional designs to investigate the relationship between the built environment (BE) and travel behaviour (TB). Since these studies cannot identify causal influence, the use of designs that provide data on multiple moments in time seems necessary. This article classifies these designs and describes how they can be applied to identify causality in this relationship. We recommend the use of natural experiments to assess the impact of changes in land use/infrastructure and prospective longitudinal designs to assess the impact of residential or job moves. In addition, the role of the BE can be explored by assessing the impact of (1) deliberate TB change experiments and (2) changes in household circumstances across different spatial contexts over time. The use of randomised experimental designs is recommended for the former and prospective longitudinal designs for the latter. The article concludes with an outlook on future research.

Residential self-selection, reverse causality and residential dissonance. A latent class transition model of interactions between the built environment, travel attitudes and travel behavior

Travel-related attitudes and dissonance between attitudes and the characteristics of the residential built environment are believed to play an important role in the effectiveness of land use policies that aim to influence travel behaviour. To date, research on the nature and directions of causality of the links between these variables has been hindered by the lack of longitudinal approaches. This paper takes such an approach by exploring how people across different population groups adjust their residential environments and attitudes over time. Two latent class transition models are used to segment a population into consonant and dissonant classes to reveal differences in their adjustment process. Interactions between (1) the distance to railway stations and travel-mode-related attitudes and (2) the distance to shopping centres and the importance of satisfaction with these distances are modelled. The models reveal mixed patterns in consonant and dissonant classes at different distances from these destinations. These patterns remain relatively stable over time. People in more dissonant classes generally do not have a higher probability of switching to more consonant classes. People adjust their built environments as well as their attitudes over time and these processes differ between classes. Implications for policies are discussed.

Causal effects of built environment characteristics on travel behaviour: a longitudinal approach

The influence of the built environment on travel behaviour and the role of intervening variables such as socio-demographics and travel-related attitudes have long been debated in the literature. To date, most empirical studies have applied cross-sectional designs to investigate their bidirectional relationships. However, these designs provide limited evidence for causality. This study represents one of the first attempts to employ a longitudinal design on these relationships. We applied cross lagged panel structural equation models to a two-wave longitudinal dataset to assess the directions and strengths of the relationships between the built environment, travel behaviour and travel-related attitudes. Results show that the residential built environment has a small but significant influence on car use and travel attitudes. In addition, the built environment influenced travel-related attitudes indicating that people tend to adjust their attitudes to their built environment. This provides some support for land use policies that aim to influence travel behaviour.

The interplay between land use, travel behaviour and attitudes: a quest for causality.

Grounding Causality in Bayesian Networks Using Qualitative Reasoning

The complexity of analysing dynamical systems often lies in the difficulty to monitor each of their dynamic properties. In this article, we use qualitative models to present an exhaustive way of representing every possible state of a given system, and combine it with Bayesian networks to integrate quantitative information and reasoning under uncertainty. The result is a combined model able to give explanations relying on expert knowledge to predict the behaviour of a system. We illustrate our approach with a deterministic model to show how the combination is done, then extend this model to integrate uncertainty and demonstrate its benefits