Data Imputation in Amos

[Gulfam Sadiq]

Hello respected members, I hope you are all fine.
I have conducted an experimental design study. After the descriptives, I had hyopthesized a structural model to test. It had 5 latent variables i.e, IDV, 2 parallel mediators followed by a serial mediator, and DV. The number of observed variables are 36. My sample size for this group was just 56. 
1. is this enough sample size to defend in case of experimental designs. because I am a student so i had limited number of resources to conduct a large scale experiment. Guide me on this subject.
2. When I draw my measurement model in AMOS, and impute the values after drawing covariances and after checking validity and reliability and model fitness, and then run my structual analysis on imputed variables, my model does not get acceptable fit. CFI and TLI values are above .95 but rmsea value is above .1. But If impute the values after drawing structural paths in my model instead of covariances and then impute the variables then my model recieves acceptable fit, CFI, TLI > .95 and rmsea < .08.. What could be possible reason for this and can I impute after drawing paths instead of covariances?
3. If we provide validity, reliability, and model fitness of measurement model, then do we need to provide model fitness again when we run analysis on imputed variables?
Regards,
Ghulfam Sadiq,
Southwest University, China

[Neeraj Kaushik]

Dear Ghulfam 

There are many things you've written in your email. Let me address them.

1. Whenever we use Regression, the sample size is 50+8k where k is no of vars

So when we have 1 IDV the min sample of 58 shd be used.

IF you've 36 IDVs then the min sample of 338 shd be required.

2. Imputation of data shd be done after the CFA. The results will differ when you impute after structural model.

3. Since you mentioned that you are a students and cudnt afford large sample, you shd reduce the no of vars you are working on.

Best wishes  

 

--
The members of this group are expected to follow the following Protocols:
1. Please search previous posts in the group before posting the question.
2. Don't write the query in someone's post. Always use the option of New topic for the new question. You can do this by writing to dataanaly...@googlegroups.com
3. It’s better to give a proper subject to your post/query. It'll help others while searching.
4. Never write Open-ended queries. This group intends to help research scholars, NOT TO WORK FOR THEM.
5. Never write words like URGENT in your posts. People will help when they are free.
6. Never upload any information about National Seminars/Conferences. Send such information
in personal emails and feel free to share any RESEARCH-related information.
7. No Happy New Year, Happy Diwali, Happy Holi, Happy Birthday, Happy Anniversary, etc. allowed in this group.
8. Asking or sharing Research Papers is NOT ALLOWED.
9. You can share your questionnaire only once.
---
You received this message because you are subscribed to the Google Groups "DataAnalysis" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dataanalysistrai...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dataanalysistraining/19fa6d6f-24a9-4d5c-ad70-332c30673437n%40googlegroups.com.

[Gulfam Sadiq]

Thank You for this comrehensive reply. 

I have one more query.

Is it necessary to give model fitness values after imputation as well? Because the model that is valid during CFA, becomes invalid when I used imputed constructs for analysis..

Regards,

Ghulfam Sadiq.

To view this discussion on the web visit https://groups.google.com/d/msgid/dataanalysistraining/CAAd%3Dc8OhqvbNMb2OVj-%3DDy33bX7i7qRW%2BVpv_nE1yuPXvqKQCA%40mail.gmail.com.

[Neeraj Kaushik]

Different people answer this question differently.

You will find papers where the authors have used the model fit indicators in the structural model too.

While others impute after CFA and perform the structural model analysis in SPSS, where there is no model foit indicator. So, it also matters which software you are using for the structural model analysis. 

To view this discussion on the web visit https://groups.google.com/d/msgid/dataanalysistraining/CAFpb-_HpZ5jEzUjxjYy9GYx9dFtQ%2BQu-4m4xUFUHosrE0Q%3DNpw%40mail.gmail.com.

[Gulfam Sadiq]

okay, I understand. Thank you for the response