Comparison of methods for dealing with missing values in the EPV-R

Area of Science:

• Psychometrics
• Criminology
• Sociology

Background:

• Assessing intimate partner violence (IPV) risk is crucial.
• The Severe Intimate Partner Violence Risk Prediction Scale-Revised (EPV-R) is a Spanish tool for this purpose.
• High rates of missing data are a common challenge in IPV risk assessment scales.

Purpose of the Study:

• To evaluate the Severe Intimate Partner Violence Risk Prediction Scale-Revised (EPV-R).
• To analyze missing data patterns in the EPV-R.
• To compare imputation methods for handling missing data in the EPV-R.

Main Methods:

• Analysis of missing data patterns in a sample of 1215 male abusers.
• Evaluation of imputation methods using simulated data mirroring empirical missingness.
• Comparison of the scale's original imputation procedure with Item Response Theory (IRT)-based imputation.

Main Results:

• The EPV-R's original imputation method yields acceptable results.
• Item Response Theory (IRT)-based imputation demonstrates higher accuracy.
• IRT imputation offers additional advantages for handling missing data in the EPV-R.

Conclusions:

• Item Response Theory (IRT) is a valuable tool for imputing missing data in IPV risk assessment questionnaires.
• IRT-based imputation enhances the accuracy and utility of scales like the EPV-R.
• This approach addresses a significant methodological challenge in violence risk assessment instruments.