Ignoring Clustering in Confirmatory Factor Analysis: Some Consequences for Model Fit and Standardized Parameter...

Area of Science:

• Multilevel modeling
• Statistical analysis
• Psychometrics

Background:

• Researchers often collect multilevel (clustered or nested) data.
• Common analysis methods include disaggregation (ignoring clustering) or aggregation (averaging micro-level units).
• These methods may introduce bias when analyzing nested data.

Purpose of the Study:

• To investigate the effects of ignoring data nesting in confirmatory factor analysis (CFA).
• To examine the bias in model fit and standardized parameter estimates.
• To compare disaggregation and aggregation approaches against multilevel CFA (MCFA).

Main Methods:

• Confirmatory factor analysis (CFA) was used to analyze multilevel data.
• The study examined the impact of disaggregation and aggregation methods.
• Model fit and standardized parameter estimates were assessed under varying intraclass correlation (ICC) and cluster sizes.

Main Results:

• Disaggregation increases model misfit, particularly with high intraclass correlation (ICC).
• Aggregation accurately detects macro-level model misfit but deviates parameter estimates.
• Standardized parameter estimates from both methods are biased, with deviations depending on ICC and cluster size.

Conclusions:

• Ignoring the nested nature of data in CFA leads to biased results.
• Disaggregation inflates model misfit, while aggregation can mask micro-level effects.
• Multilevel CFA (MCFA) or alternative multilevel methods are recommended for accurate analysis of clustered data.