A Latent Class Extension of Signal Detection Theory, with Applications

Area of Science:

• Psychology
• Statistics
• Machine Learning

Background:

• Signal detection theory (SDT) is a framework for analyzing observer performance.
• Existing SDT models may not adequately capture complex categorical detection tasks.
• There is a need for methods to classify cases based on observer performance.

Purpose of the Study:

• To present a latent class extension of signal detection theory.
• To illustrate applications in detecting latent categorical events and classifying cases.
• To provide a flexible model for analyzing observer performance.

Main Methods:

• Developed a latent class extension of signal detection theory.
• Extended the model to accommodate more than two latent classes.
• Provided sample programs for fitting models using latent class analysis and structural equation modeling software.

Main Results:

• The latent class extension effectively summarizes observer performance using detection and response criteria.
• The model allows for the analysis of more complex categorical detection scenarios.
• The approach facilitates the selection and classification of cases based on performance.

Conclusions:

• The latent class extension of signal detection theory provides a powerful tool for analyzing observer performance in categorical detection tasks.
• This approach has implications for rater training and the validation of classifications.
• The model offers a flexible and extensible framework for various applications.