The Psychometric Latent Agreement Model (PLAM) for Discrete Latent Variables Measured by Multiple Items.

Area of Science:

• Psychometrics
• Statistical Modeling
• Behavioral Science

Background:

• Accurate subpopulation membership estimation is crucial for decisions like hiring and performance evaluations.
• Existing methods may struggle with indeterminate classifications due to insufficient rating information.
• Latent trait variables are fundamental in understanding unobserved individual characteristics.

Purpose of the Study:

• To introduce the Psychometric Latent Agreement Model (PLAM) for estimating subpopulation membership.
• To incorporate a latent Type variable to handle indeterminate classifications.
• To estimate agreements among discrete latent trait variables.

Main Methods:

• Development of the Psychometric Latent Agreement Model (PLAM).
• Introduction of a binary latent Type variable to identify indeterminate cases.
• Application of PLAM to real-world behavioral rating data from single and multiple rater types.

Main Results:

• PLAM successfully estimates subpopulation membership (e.g., satisfactory vs. unsatisfactory performers).
• The latent Type variable effectively separates reliably classified individuals from those with indeterminate status.
• Agreements among discrete latent trait variables were successfully estimated in empirical examples.

Conclusions:

• PLAM offers a robust framework for classifying individuals based on latent traits, accounting for classification uncertainty.
• The model has practical implications for objective decision-making in personnel selection and performance assessment.
• Behavioral ratings can be effectively utilized for determining qualified/unqualified or pass/fail groupings.