In a specific CFA model, after running, I cannot view the path coefficients in the graphical representation. The button seems to be disabled and appears to be grey. This has not happened in any other models. Help would be hugely appreciated.
Sorry for waiting so long. I thought about your problem. Here is one of the solutions:
https://youtu.be/B7YOv7hSohY
Check the factor loadings as shown in the video.
If it does not work, move the constraints (1 on a path from factor to one of the items) to another item. I don't know why, but it works sometimes.
If it doesn't work, check your data. Try to delete factors one by one to find out, which of them is casing the problem (and always save your new file! )
One more comment: we usually use at least three items for a factor. If you find a good literature why is it allowed to use two items - send me the paper, I would appreciate it.
It means that the model did not run. The path coefficients are shown if everything runs perfectly.
1. Go to you results table. Assumably, you will find a lot of data missing. Look in regression weights. The place where the program stopped could indicate the problem.
2. Usually, however, the problem lies in missed error terms, constraints, wrong data and so on. If you send me the model I can check it. You can just attach it to your next message here.
thanks for the advice. I did not realize that it did not run - because I was able to see the coefficients in the Text Output and did not get an error message. I cannot seem to find the reason though. It would be fantastic if you could have a look. Thanks so much! The AMOS Data is attached.
The issue is that your iteration limit is reached. In the output tab you can increase the iteration limit in the output tab. You may have some negative variances so check on that.
Sorry for waiting so long. I thought about your problem. Here is one of the solutions:
https://youtu.be/B7YOv7hSohY
Check the factor loadings as shown in the video.
If it does not work, move the constraints (1 on a path from factor to one of the items) to another item. I don't know why, but it works sometimes.
If it doesn't work, check your data. Try to delete factors one by one to find out, which of them is casing the problem (and always save your new file! )
One more comment: we usually use at least three items for a factor. If you find a good literature why is it allowed to use two items - send me the paper, I would appreciate it.
I would add that you're model is probably locally unidentified and one or several of the two indicator latents do not correlate with any of the other latents. In this case, you coud set an equality constraint at the loadings, moving this set of indicators towards a tau equivalent model.
As for the two-indicator-model per se: Of course it would be wonderful if we could identify or create 4-5 or even more indicators that are really homogenous. However, in practice this is seldom plausible as changes in item wordings result in causal heterogeneity (i.e., the items measure different latents). This is the worst thing that can happen, as it violates the basic structure of the respective measurement model.
I was heavily influenced by the work of Leslie Hayduk which recommends using the 2 (or even 1) best - that is - theoretically clearest indicators. You can search for his articles.
One article that hits the target is
Hayduk, L. A., & Littvay, L. (2012). Should researchers use single indicators, best indicators, or multiple indicators. BMC Medical Research Methodology, 12(159), 1–17. doi:10.1186/1471-2288-12-159
Then I go to Analysis property, numerical and change the " iteration limit " from 50 to 100. Then it works normally, all path coefficients appear in the graph diagram.
That is the problem with proprietary software. We need to follow what it is, it never does what we want. Use either R or Onyx you can get diagrams with all estimates.
I know I'm responding very late to this, but I just had the same problem. I ran the analysis, it seemed to run okay, but the button to show the output was greyed out and I couldn't see any results.
After much searching around and playing around with the model, I decided to start again from scratch. I resaved the data and created a new model. It worked. I have no idea what when wrong, but it all worked out in the end.