Dear Colleagues,
I recently had an issue while running CFA. I have four single indicator factors which will be used to form formative constructs. Before doing so, I want to test the validities of the measures. My research design is to first ask the respondents to pick the situations they have experience, then based on the selected situations, they rate the four measures from 1-7. In this case, my database have quite a few "missing values" (respondents answered the questions based on different situations, not picked situations will have missing values in the four questions).
May I ask how I can conduct CFA or test the validities of the four single indicator measures in this situation? I doubt treating them as normal missing values is a appropriate way. Can I replace them with zero but when the zero here is different from the quantitatively meaningful zero.
Thank you in advance for any inspiring opinions.
Hello, Yang Liu
I think this discussion is useful,
https://www.researchgate.net/post/When_imputing_missing_values_in_SPSS_how_many_mean_or_meadian_of_nearby_points_is_recommended
Good lock
لمناقشة هذا السؤال يجب الاخذ بنظر الاعتبار القيم المفقودة عند اختبار صلاحية الإجراءات
Hello Yang,
I'd first focus on the relationships (observed or estimated) for individual scenarios, as you would have "complete" data for those instances.
Then, you could ask the question as to whether the observed relationships in one scenario were congruent with those from any other(s), by comparing the matrices of observed/estimated correlations. You can do this by inspection, or by tests of equality (for a set of correlations). Technically, the relationships are likely to be more so dependent than independent, which admittedly makes the comparison more challenging.
I doubt that imputation makes a lot of sense, if scenario choice in any way influences the perceptions of your respondents.
Good luck with your work.
You did not mention it, but I assume that you are doing it in IBM SPSS AMOS. As per my limited erudition, you need to conduct a separate EFA before conducting CFA. The assumptions of EFA will guide you about the possibility of CFA. For this purpose, you can conduct EFA by selecting the one situation one by one and the related items.
If, you got it. Then I could explain the next step?
- Badges
- Science topic
- Similar topics
- Social Science
- Media
- Journalism