Event History Analysis for Psychological Time-to-Event Data

Speaker: Sven Panis, Swiss Federal Institute of Technology Zurich, CH
Series: CBU Monday Methods Meeting

Title: Event History Analysis for Psychological Time-to-Event Data

Abstract: Characterizing the cognitive states that unfold within a trial requires measuring behavioral performance with high temporal resolution. Conventional methods, such as means, quantile plots, and delta plots, do not capture how the probability of a response changes over time given that no response has yet occurred, and they cannot handle right-censored observations (e.g., trials without a response) or time-varying predictors. Here we explain how event history analysis, an established technique in biomedicine and engineering, can be applied to psychological time-to-event data, including response times, saccade latencies, fixation durations, time to maximum curvature in movement trajectories, and percept durations in bistable perception tasks. For choice data, we combine hazard functions with conditional accuracy functions, which track the probability that a response is correct given that it occurs in a particular time bin. We cover the definitions of hazard, survivor, and conditional accuracy functions, provide code for computing descriptive statistics, and show how to fit discrete-time hazard and conditional accuracy models within both frequentist and Bayesian frameworks. Empirical examples with choice response time data illustrate how this approach reveals the within-trial dynamics of cognitive processes that conventional analyses obscure.

Speaker: Sven Panis, Swiss Federal Institute of Technology Zurich, CH

Bio: Dr. Sven Panis is an experimental psychologist who joined the lab of Dr. Richard Ramsey at ETH Zürich in January 2024. His research focuses on the temporal dynamics of selection, suppression, and decision processes. Using event history (hazard) analysis, he examines how behavior unfolds over time within trials in standard cognitive paradigms, and what these dynamics imply for cognitive theory. Sven obtained his PhD in Experimental Psychology from KU Leuven in 2008 and has led his own research projects funded by FWO-Flanders (2009–2012, Leuven, Belgium) and the DFG (2019–2022, Kaiserslautern, Germany). He also has experience with fMRI, eye-tracking, and pupillometry. Examples of hazard functions estimated from empirical RT data are available at https://sven-panis.github.io.

Venue: MRC CBU West Wing Seminar Room and Zoom (; )