Replication using confirmatory latent class analysis: A school climate example.

Area of Science:

• Psychology
• Educational Psychology
• Quantitative Psychology

Background:

• Large datasets and computation software have increased mixture modeling use, including latent class analysis (LCA).
• Secondary data analysis using LCA is common in school psychology, but results can be dataset-specific.
• Replication guidance for LCA is needed to ensure findings reflect stable patterns.

Purpose of the Study:

• To provide guidance for shifting from exploratory LCA to confirmatory LCA (CLCA).
• To confirm previous exploratory LCA findings on student school climate using CLCA.
• To demonstrate the stability of identified response patterns across different data collection periods.

Main Methods:

• Utilized confirmatory latent class analysis (CLCA) to scrutinize prior exploratory LCA results.
• Applied CLCA to student school climate perception data (n=24,051) from the 2017-2018 academic year.
• Compared CLCA findings with an earlier LCA analysis of 2016-2017 data.

Main Results:

• Confirmatory LCA results indicated substantial similarity in response patterns compared to prior exploratory analysis.
• The study confirmed a five-class solution for student school climate experiences.
• The five-class solution demonstrated replicability across different academic years and respondent groups.

Conclusions:

• The five-class solution for student school climate experiences is stable and replicable.
• Confirmatory LCA supports the validity of identified patterns beyond dataset-specific artifacts.
• Guidance is provided for researchers using nascent CLCA methods to ensure robust findings.