A nonlinear regression approach to estimating signal detection models for rating data

Area of Science:

• Psychology
• Statistics
• Cognitive Science

Background:

• Traditional signal detection theory (SDT) analysis often requires specialized software for parameter estimation.
• Analyzing rating data in SDT can be complex, limiting covariate exploration.
• Current methods may lack efficiency and flexibility in statistical analysis.

Purpose of the Study:

• To present a novel regression-based approach for estimating signal detection parameters from rating data.
• To demonstrate the efficiency and utility of this method using standard statistical software.
• To facilitate the investigation of covariate effects on SDT parameters.

Main Methods:

• The study employs statistical modeling of ordinal data using a regression framework.
• The methodology is implemented using widely available statistical software, such as SAS.
• The approach was validated on perceptual task data and recognition memory rating data.

Main Results:

• The proposed regression approach provides an efficient alternative to specialized software for SDT parameter estimation.
• This method allows for straightforward exploration of how covariates influence model parameters.
• The approach is effective for both simple and complex experimental designs involving rating data.

Conclusions:

• A regression-based method offers a more efficient and accessible way to estimate signal detection parameters from rating data.
• This approach enhances the ability to study factors influencing decision-making processes.
• The methodology integrates seamlessly with standard statistical practices, broadening its applicability.