A Bayesian latent group analysis for detecting poor effort in a sample of cognitively impaired patients

Abstract

Using a Bayesian latent group analysis in a simulation design, we recently showed a high diagnostic accuracy when assessing effort in the context of malingered memory deficits. We here further evaluate our Bayesian model in a sample of cognitively impaired patients. The main analysis showed both high sensitivity and specificity, thus corroborating a high diagnostic accuracy of the model. Additional analysis showed variations on effort estimates after changes in malingering base rates. Variations affected sensitivity, but not specificity, which is in line with typical findings in malingering research. These data suggest that Bayesian analyses may complement and improve existing effort measures.

Keywords:

• Malingering
• Poor effort
• Bayesian analysis
• Base rates
• Cognitive impairment

We thank our colleagues in the Physiological Psychology Department and the Center of Excellence “Cognitive Interaction Technology” (CITEC) Research Groups at Bielefeld University for their technical support and helpful advice. We would like to thank Simone Horstmann, patients, and staff of the Evangelic Hospital of Bielefeld (Evangelisches Krankenhaus Bielefeld) for their help and assistance when carrying out our clinical study. We would also like give special thanks to David T. R. Berry for kindly assisting us while preparing our manuscript.

This work was supported by the German Research Foundation (Deutsche Forschungsgemeinschaft) [grant number EC 277 to M.P., H.J.M.); Becas Chile Scholarship, CONICYT (A.O.).