Reaction times (RTs) are a key measure in experimental psychology, reflecting the time taken to respond to stimuli. Classical methods for analyzing RT data focus on aggregated data, but mixed-effects modeling allows for more detailed analysis of individual responses. This study advocates for flexible approaches to RT transformation, minimal data trimming, and model criticism. It also highlights the importance of accounting for trial-to-trial dependencies and interactions involving fixed-effect factors. RT distributions are often skewed, and outliers can significantly affect results. The study shows that the Inverse-Gaussian distribution fits RT data well, but the best model depends on the experimental context. Outlier handling requires careful consideration, as extreme values can distort results. The study also addresses temporal dependencies, showing that trial-by-trial correlations can be modeled using covariates like trial number and preceding RT. Mixed-effects models allow for the inclusion of multiple random-effect factors, such as subjects, items, and poems, providing a more comprehensive analysis. The study illustrates these methods using a large dataset of self-paced reading latencies, showing how fixed and random effects contribute to RT variability. The results highlight the importance of considering both individual and contextual factors in RT analysis.Reaction times (RTs) are a key measure in experimental psychology, reflecting the time taken to respond to stimuli. Classical methods for analyzing RT data focus on aggregated data, but mixed-effects modeling allows for more detailed analysis of individual responses. This study advocates for flexible approaches to RT transformation, minimal data trimming, and model criticism. It also highlights the importance of accounting for trial-to-trial dependencies and interactions involving fixed-effect factors. RT distributions are often skewed, and outliers can significantly affect results. The study shows that the Inverse-Gaussian distribution fits RT data well, but the best model depends on the experimental context. Outlier handling requires careful consideration, as extreme values can distort results. The study also addresses temporal dependencies, showing that trial-by-trial correlations can be modeled using covariates like trial number and preceding RT. Mixed-effects models allow for the inclusion of multiple random-effect factors, such as subjects, items, and poems, providing a more comprehensive analysis. The study illustrates these methods using a large dataset of self-paced reading latencies, showing how fixed and random effects contribute to RT variability. The results highlight the importance of considering both individual and contextual factors in RT analysis.