I'm having difficulty achieving discriminant validity for my model in AMOS. I have collected data for the same variables from the same participants across four weeks. In the CFA for week 1, everything was fine. In week 2, however, the square root of the AVE for one factor is lower than its correlation with another factor: 0.747 < 0.755. Is a difference of 0.008 acceptable? I have tried removing the items with the lowest factor loadings, but they tend to make the situation worse. Is it reasonable to just report the small difference and acknowledge that discriminant validity wasn't achieved in week 2?
That is what most people would do. Acknowledge the weakness and move on. You see it with measurement equivalence/invariance when people are using measures in different samples. They note that they did not achieve equivalence and then go on to interpret results of the substantive questions their study is addressing.
Paul E. Spector Thanks for your reply. That's what I'm leaning towards doing.
It reasonable to report and aknowledge the little difference. But you mus try to make hypotheses on the reasons of this difference in your discussion. Does it link to the items and their comprehension by the participants? Does it link to the limitee time between the collection of data rounds?...
You may see at https://stats4edu.com/services/structural-equation-modeling/
- Badges
- Science topic
- Similar topics
- Psychology