The invariance assumption in process-dissociation models: an evaluation across three domains

Area of Science:

• Cognitive Psychology
• Mathematical Psychology

Background:

• Process-dissociation models, a type of multinomial processing-tree model, are well-established in experimental psychology.
• Debates have refined these models, but empirically testing the invariance assumption remains challenging.
• Violations of invariance can lead to biased analyses and incorrect theoretical conclusions.

Purpose of the Study:

• To develop and apply novel methods for empirically evaluating the invariance assumption in process-dissociation models.
• To investigate the extent of invariance assumption violations in prominent psychological tasks.

Main Methods:

• Utilized recent advancements in multinomial modeling.
• Applied new empirical approaches to examine invariance in six studies.
• Focused on three application domains: the Stroop task, cued recall (recollection vs. habit), and racial bias in the weapon task.

Main Results:

• The invariance assumption was found to be violated to a considerable extent across all investigated domains.
• Empirical examination confirmed widespread issues with the dominant process invariance.

Conclusions:

• The findings highlight significant limitations of conventional process-dissociation analyses due to violated invariance.
• New methods offer a way to empirically assess and potentially correct for these violations.
• Reinterpreting results from the Stroop task, cued recall, and weapon task is necessary.