Problems when fixing the response bias parameter z in drift diffusion analysis : A Commentary on Stafford et al....

Area of Science:

• Cognitive Psychology
• Computational Neuroscience
• Psychometrics

Background:

• The Drift Diffusion Model (DDM) is a prominent tool for analyzing decision-making processes, particularly in tasks involving speed-accuracy trade-offs.
• Stafford et al. (2020) developed an online power analysis tool based on the DDM, which is valuable for experimental design.
• The EZ method was used by Stafford et al. for DDM parameter estimation in their simulations.

Purpose of the Study:

• To evaluate the accuracy of the EZ method for Drift Diffusion Model (DDM) parameter estimation.
• To identify potential biases in DDM parameter estimates when response bias is not equal to 0.5.
• To inform researchers about limitations of existing DDM power analysis tools.

Main Methods:

• Simulation study analyzing the Drift Diffusion Model (DDM).
• Application of the EZ method for parameter estimation.
• Comparison of EZ method bias with maximum likelihood estimation (MLE) with fixed response bias.
• Investigation of estimation bias under varying true response bias conditions.

Main Results:

• The EZ method introduces severe estimation bias for all DDM parameters when the true response bias (z) is not 0.5.
• Bias patterns generated by the EZ method differ from those produced by maximum likelihood estimation with z fixed at 0.5.
• The EZ method's inability to estimate the response bias parameter is a key source of the observed estimation bias.

Conclusions:

• The EZ method is not suitable for DDM parameter estimation when response bias deviates from 0.5.
• Researchers using DDM-based power analysis tools must account for potential response bias.
• Alternative estimation methods or careful consideration of the EZ method's limitations are necessary for accurate DDM analysis and power calculations.