I conducted my CFA analysis in R. Does anyone know how to get discriminant and convergent validity for CFA? I watched James Gaskin's videos on validity, but he uses output from AMOS. Could I transfer my output in R and use his StatsTools to get discriminant and convergent validity for CFA?
I agree with Matthew Kerry that this typically just involves correlation coefficients, where you specify other items where you expect a positive correlation (convergent validity) or null-negative correlations (discriminant validity).
Does this discussion help?
https://stats.stackexchange.com/questions/20196/which-package-to-use-for-convergent-and-discriminant-validity-in-r
Hi Yang,
You can use the results of the CFA to test for convergent and discriminant validity. First you need to establish uni-dimensionality. The uni-dimensionality assumption is supported if the CFA demonstrates high item loadings and acceptable model fit indices.After uni-dimensionality has been achieved, statistically significant factor loadings are considered to be evidence for convergent validity.
Discriminant validity is established for the scales and the items. You can test construct level discriminant validity using the comparison of average variance
extracted to shared variance (AVE-SV criterion) proposed by Fornell and Larcker
(1981). According to the AVE-SV criterion, discriminant validity for a construct exists when the construct’s squared root of average variance extracted (AVE) is greater than its correlations with all other constructs in the model. At item-level for establishing discriminant validity, you can use the CFA results
and see if there is any potential cross-loadings. No noticeable cross loading suggests the availability of discriminant validity at item-level.
Hope this would help.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R