Evidence Accumulation Model Selector

Choosing an evidence accumulation model depends on trial counts and bias considerations. Use EZ-diffusion for low trial counts, full DDM for complete RT distributions with accuracy data, and LBA or race models for multi-alternative experimental designs.

You need an evidence accumulation model when analyzing choice-response time data from experiments with multiple response options. These models decompose full RT distributions and accuracy data to estimate underlying cognitive decision-making parameters.

Fitting a full DDM or LBA model requires sufficient trial counts per condition to ensure reliable parameter estimation. You also need access to full RT distributions and accuracy data, plus familiarity with evaluation metrics like WAIC, BIC, or Bayes factors.

To evaluate and compare evidence accumulation model fits, use model-fitting evaluation metrics such as WAIC, BIC, and Bayes factors. This structured decision framework also guides parameter recovery checks during the model selection process.

You should not use the full DDM when your experimental design involves multiple response options better suited to LBA or race models, or when low trial counts per condition necessitate the simpler EZ-diffusion approach.