Scale validation in applied health research: tutorial for a 6-step R-based psychometrics protocol

Area of Science:

• Health Sciences
• Psychometrics
• Statistical Modeling

Background:

• Applied health research frequently uses multi-item scales for measurement.
• Traditional psychometric validation is often separate from substantive studies.
• Methodologists advocate for integrated psychometric analysis within applied research contexts.

Purpose of the Study:

• To introduce a 6-step protocol for comprehensive psychometric analysis in R.
• To enable detailed diagnosis of scale properties like structural validity and internal consistency.
• To facilitate the integration of psychometric validation into substantive health research.

Main Methods:

• A 6-step protocol for diagnosing psychometric properties of ordinal and binary scales.
• Utilizes item response theory, factor analysis, and classical test theory.
• Demonstrated using the Roland Scale (RM-SIP) in a chronic pain cohort with an R Markdown script.

Main Results:

• 15 of 24 RM-SIP items formed a unidimensional ordinal scale.
• Demonstrated good homogeneity (H=0.43) and reliability (alpha=0.86, omega=0.87).
• Regression models showed comparable results for predicting disability.

Conclusions:

• Psychometric properties can be effectively assessed within substantive studies.
• The protocol and R script aid in measure improvement and validation.
• Applied researchers can adapt this approach for robust questionnaire analysis.