I want to use SPSS Amos to calculate SEM because I use SPSS for my statistical analysis. I have already found some workarounds, but they are not useful for me. For example, using a correlation matrix where the weights are already applied seems way too confusing to me and is really error prone since I have a large dataset. I already thought about using Lavaan with SPSS, because I read somewhere that you can apply weights in the syntax in Lavaan. But I don't know if this is true and if it will work with SPSS. Furthermore, to be honest, I'm not too keen on learning another syntax again.
So I hope I'm not the first person who has problems adding weights in Amos (or SEM in general) - if you have any ideas or workarounds I'll be forever grateful! :)
Well, I would suggest weighting the data in SPSS and then directly using the weighted data in Amos. This avoids the complexity of replicating cases or aggregating data manually. (I consider this approach the easiest)
Step 1 in SPSS:
- Open SPSS and load your dataset (e.g., your_data.sav).
- Use the following SPSS syntax to apply weights to your data:
GET FILE='your_data.sav'.
WEIGHT BY weight_variable.
- This tells SPSS to use the specified weight variable in all analyses, including when saving the data.
- Save the weighted dataset:
SAVE OUTFILE='weighted_data.sav'.
Step 2: Use the Weighted Data in Amos
- Start Amos and load the dataset weighted_data.sav by clicking on the file menu and selecting your saved file.
- Create your SEM model by drawing the variables and specifying the paths as needed. (Use Amos’s graphical interface to define the relationships between your observed and latent variables.)
- Once your model is specified, click on the 'Calculate Estimates' button to run the analysis. Amos will use the weighted data you prepared in SPSS.
I think this is very straightforward and clear (in comparison with using Lavaan). Good luck!
Ayoub Zrari thank you very much for your reply!
I just realized that my original question was not specific enough...
So the point of my problem is that Amos can't apply case weights. If I do what you suggested, then I get the following error from Amos: "Amos will ignore the case weights in the file. Each case will be given a weight of 1. This will affect the results of your analysis". And this is the point of my whole dilemma, because of this I need some workaround. Do you have any other advice? IBM is also aware of this: https://www.ibm.com/support/pages/amos-wont-use-case-weights-spss but I need to apply sample weighting.
Thanks a lot!
I see. Have you thought about using R Essentials integration with SPSS? I believe this will allow you to use the familiar SPSS environment while taking advantage of Lavaan's capabilities for SEM with weights.
Yes, I already tried that, but I prefer Amos and I don't know if it will work to apply the weights in Lavaan. Guess I'll have to try. Thanks for your help!
- Badges
- Science topic
- Similar topics
- Organizational Psychology
- Training