Accounting for time-dependent covariates in driving simulator studies

Driving involves multiple cognitive processes that are influenced by a dynamic external environment and internal feedback loops. These processes are typically studied in a simulator environment to capture time-dependent driver performance measures. The primary objective of this research is to show that data analysis techniques that ignore or improperly treat time-dependent covariates will lead to erroneous estimates and conclusions. This is demonstrated with a driving simulator study that was used to test whether a significant decrease in performance occurs in the presence of auditory and visual distractions. A total of 28 drivers participated in a 2 (age)× 7 (strategy) repeated measures experiment. The response variable—accelerator release time—was analysed with and without consideration of time-dependent covariates. Using the inverse headway distance as a time-dependent covariate corrected logically inconsistent results obtained when the covariate was ignored. This indicates that ignoring covariates can actually lead to inappropriate design or policy implications.