Higher-order factor analysis of an instrument with dichotomous data

Area of Science:

• Health Services Research
• Psychometrics
• Nursing Practice

Background:

• Community acceptance is crucial for integrating advanced practice providers like nurse practitioners (NPs) and physician assistants (PAs).
• Measuring community acceptance often involves instruments with mixed data types (dichotomous and Likert scale items).
• Traditional factor analysis methods face challenges with dichotomous data, potentially compromising validity.

Purpose of the Study:

• To employ higher-order factor analysis to validate an instrument measuring community acceptance of NPs and PAs.
• To address the methodological controversy of using factor analysis with dichotomous data.
• To identify the underlying dimensions of community acceptance in rural settings.

Main Methods:

• Utilized higher-order factor analysis, including maximum likelihood extraction and oblique rotation.
• Factored the correlation matrix to derive second-order factors.
• Interpreted second-order factors using factor pattern matrices and item correlations.

Main Results:

• Higher-order factor analysis successfully identified underlying dimensions of community acceptance.
• The analysis provided factorial validity for concepts including knowledge, competence, access, and trust.
• This method offers a robust approach for instruments with mixed-format items.

Conclusions:

• Higher-order factor analysis is a suitable method for validating instruments with dichotomous and Likert scale items.
• The study confirmed key dimensions influencing community acceptance of NPs and PAs in rural areas.
• Findings support the reliable measurement of community perceptions regarding advanced practice providers.