Correlation between residuals in confirmatory factor analysis: a brief guide to their use and interpretation
Introduction: The inclusion of correlations between residuals in measurement models is a common practice in psychometric research and is predominantly oriented to the statistical improvement of the model through increase (for example, IFC) or decrease (for example, RMSEA) of the magnitude of certain...
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Instituto Peruano de Orientación Psicológica – IPOPS
2019-09-01
|
| Series: | Interacciones: Revista de Avances en Psicología |
| Subjects: | |
| Online Access: | http://dx.doi.org/10.24016/2019.v5n3.207 |
| Summary: | Introduction: The inclusion of correlations between residuals in measurement models is a common practice in psychometric research and is predominantly oriented to the statistical improvement of the model through increase (for example, IFC) or decrease (for example, RMSEA) of the magnitude of certain adjustment indices, rather than understanding the nature of these associations. This methodological report aims to present to the reader the modeling, management, and interpretation of correlated residuals in a framework of confirmatory factor analysis and poor specifications. Method: Using data from a previously presented study of 521 psychology students at a private university in Metropolitan Lima (75.8% women). The Flowering Scale is used to perform the analyses. Results and Discussion: These specifications would not have a real impact on the relationship of the elements with the construct they evaluate, so they do not contribute modifications to the understanding of the model. Therefore, specifying correlations between residuals could mask a poorly specified model, or with internal failures, by increasing spurious adjustment rates. |
|---|---|
| ISSN: | 2411-5940 2413-4465 |