When accuracy rates and mean response times lead to false conclusions: A simulation study based on the diffusion model

Abstract

The diffusion model belongs to the class of sequential sampling models. Based on information about RT distributions from binary tasks, the model allows the separation of distinct processes underlying decision-making. With the present simulation study, we demonstrate that diffusion modeling has essential advantages over the mere analysis of behavioral variables such as accuracy rate or mean RT. Specifically, our results show that qualitatively different sets of diffusion model parameters can lead to the same pattern of these aggregated variables, which renders the interpretation of these measures ambiguous. Even apparently clear patterns of results can be based on completely different parameter sets, which reflect completely different cognitive processes. For example, if one group\textbackslash{}IeC {\textbackslash{}textquoteright }s performance (or performance in one experimental condition) is superior in both mean RT and accuracy rate (i.e., fast responses and few errors), this does not necessarily mean that in this group (or condition) information processing is faster (drift rate of the diffusion model). Accordingly, we conclude that the mere analysis of behavioral variables can result in false conclusions.