This is very vey helpful and informative video. Thanks for all of your obvious efforts you put in these videos. Could you please let us know from where did you get the table (Assessing the model fit and the cut-off points)? I can see the reference but which page of the book and what exact edition?
Pleasure. I am glad you liked it. You may refer to Meyers, L. S., Gamst, G., & Guarino, A. J. (2016). Applied multivariate research: Design and interpretation. Sage publications.
Excellent explanation and knowledge shared, my question is even if after doing all this the model fitness does not improve, what should be done? Also how many minimum items to have for a construct?
Pleasure. I am glad you liked it. Even after doing everything, the results do not improve, there are issues with your data. You should have at least three items left in each construct, but make sure the content validity is fine.
I noticed for standardised residual covariance you removed sl3, may i know why? what is the range of numbers should the matrix reflect to show a good fit? Also, I would like to know what does covariance between the error terms mean.
Salam. What is a cut-off for MI? In a handbook on SEM 2nd Edition by Zainudin Awang, it suggests for the modification if, MI >15. I am rather confused because, i correlated the errors which had MI >10. In one of the first-order constructs of Higher-order model, the correlation between errors is 0.2 - 0.3. Standardized Residual Covariances also seem acceptable. Shall i delete one of the indicators?
Sir, thank you for making wonderful videos. As i have collected my data and going to use AMOS, is it necessary to first of all check the normality of IDVs and DVs?
Sir, i didnt get modek fit values in the text output and I got a error massage as Iteration limit reached The results that follow are therefore incorrect. The model is probably unidentified. In order to achieve identifiability, it will probably be necessary to impose 40 additional constraints. What should I need to do to correct this issue please help me
You can simply report this as part of measurement model assessment. Write that in order to improve model fit, error terms of the same construct were covaried based on higher modification indices
Respected Sir, I have watched full vedio, plz guid whether we should use this modified CFA data file for structural model and also for meditation analysis. Or after modification of structural model this modified structural data file will be used for mediation.plz your response will be great solution for me like your previous responses.
Sir, I want to test a model via SEM where one of my moderators will be a categorical variable. So, for assessing model fit, is there any way to include the categorical variable and assess model fit? Or, the categorical moderator can only be added in the structural model? Please help me with this Sir. Thank You so much
To add clarity , I did add the categorical moderator (gender) and the model fit was fine. But how do i segregate gender into male and female in the measurement model itself? Should I do it just like we do in structural path? Or is there some other way?
Sir, i have a genuine question. If we do modifications of error terms for model fit, will we also modify those error terms in the main dataset of structural equation modeling? (if we use the same data along with some new responses)
@@researchwithfawad Sorry, I meant would we use the same CFA model (inclusive of error covariance) to test the structural model? If yes, what if we get Iteration issues? Are we allowed to remove error covariance from the structural model then? Also you have not uploaded any video to interpret goodness-of-fit measures for structural models. These are some of the confusions that are unanswered otherwise you have helped me through your videos and quick response. The way you respond back and resolve your subscriber's questions is just amazing.
It could be due to variety of reasons. You may have a reasonable sample size with simple model. It is already should perfect fit. This has to do more with justification than improvement.
Type the following in Google Scholar to find relevant papers allintext:"Perfect fit" "RMSEA=0.000" "AMOS" or allintext:"Perfect fit" "RMSEA = 0.000" "AMOS"
Your explanation such amazing sir great thank you so much
You are most welcome
Thanks for your complete explanation 🙏
My pleasure
Thank you for such comprehensive explanation. Every single lecture is helping me much more in very less time.
Thanks. I am glad you liked it
Absolutely perfect sir,very nicely explained each and every thing, it helped a lot sir
Glad to hear that
Sir, you are a great source of knowledge. You have explained different aspects nicely. Thank you for sharing such an amazing tutorial.
You are most welcome
Brilliant!! Just Brilliant!! Excellently explained with perfect examples. May Allah bless.
Many many thanks. I am glad you liked it.
This is a really amazing explanation prof. you are a great lecturer
Glad it was helpful!
This is very vey helpful and informative video. Thanks for all of your obvious efforts you put in these videos. Could you please let us know from where did you get the table (Assessing the model fit and the cut-off points)? I can see the reference but which page of the book and what exact edition?
Please refer to
Collier, J. (2020). Applied structural equation modeling using AMOS: Basic to advanced techniques. Routledge.
Thanks for the good explanation. What reference we can use to report the covariance of the error terms?
Pleasure. I am glad you liked it. You may refer to
Meyers, L. S., Gamst, G., & Guarino, A. J. (2016). Applied multivariate research: Design and interpretation. Sage publications.
Thanks a lot! This helped but my data is not normally distributed. Do I go through these same steps in this case?
Yes, AMOS is robust to minor violations of normality. But it shouldn't be too severe.
18:55 is the process of selecting MI value based on highest first, second highest next and so on.
Yes.
Excellent explanation and knowledge shared, my question is even if after doing all this the model fitness does not improve, what should be done? Also how many minimum items to have for a construct?
Pleasure. I am glad you liked it. Even after doing everything, the results do not improve, there are issues with your data. You should have at least three items left in each construct, but make sure the content validity is fine.
I noticed for standardised residual covariance you removed sl3, may i know why? what is the range of numbers should the matrix reflect to show a good fit? Also, I would like to know what does covariance between the error terms mean.
Thanks for watching. You may delete items that have values outside +/-2.
Salam. What is a cut-off for MI? In a handbook on SEM 2nd Edition by Zainudin Awang, it suggests for the modification if, MI >15. I am rather confused because, i correlated the errors which had MI >10. In one of the first-order constructs of Higher-order model, the correlation between errors is 0.2 - 0.3. Standardized Residual Covariances also seem acceptable. Shall i delete one of the indicators?
Please follow Prof. Awang.
Sir, thank you for making wonderful videos. As i have collected my data and going to use AMOS, is it necessary to first of all check the normality of IDVs and DVs?
Thanks. I am glad you liked it. Check for Skewness and Kurtosis. They should not be too high.
Sir, i didnt get modek fit values in the text output and I got a error massage as Iteration limit reached
The results that follow are therefore incorrect.
The model is probably unidentified. In order to achieve identifiability, it will probably be necessary to impose 40 additional constraints.
What should I need to do to correct this issue please help me
Please see if you have fixed at least 1 parameter to 1 in each construct.
What if everything in the estimates is below .5?
This means the fit is not good and you may need to collect more data, clean existing data.
Must I report all model fits in my paper or report those (such as CMIN/DF, RMSEA, GFI, AGFI, RMR, PGFI, etc that meet the threshold
No need to report all. You may only report CMIN/df, CFI, TLI, SRMR, RMSEA.
How to report that error terms have been covaried in order to improve model fit
You can simply report this as part of measurement model assessment. Write that in order to improve model fit, error terms of the same construct were covaried based on higher modification indices
Respected Sir, I have watched full vedio, plz guid whether we should use this modified CFA data file for structural model and also for meditation analysis. Or after modification of structural model this modified structural data file will be used for mediation.plz your response will be great solution for me like your previous responses.
Use modified one, remove any variables that have been excluded as part of initial cfa.
thankyou so much sir
Most welcome
Sir, I want to test a model via SEM where one of my moderators will be a categorical variable. So, for assessing model fit, is there any way to include the categorical variable and assess model fit? Or, the categorical moderator can only be added in the structural model? Please help me with this Sir. Thank You so much
To add clarity , I did add the categorical moderator (gender) and the model fit was fine. But how do i segregate gender into male and female in the measurement model itself? Should I do it just like we do in structural path? Or is there some other way?
Please remove the categorical moderator.
Noted Sir. Thank You once again
Sir, i have a genuine question. If we do modifications of error terms for model fit, will we also modify those error terms in the main dataset of structural equation modeling? (if we use the same data along with some new responses)
In the model only, how can the dataset be modified?
@@researchwithfawad Sorry, I meant would we use the same CFA model (inclusive of error covariance) to test the structural model?
If yes, what if we get Iteration issues? Are we allowed to remove error covariance from the structural model then? Also you have not uploaded any video to interpret goodness-of-fit measures for structural models. These are some of the confusions that are unanswered otherwise you have helped me through your videos and quick response. The way you respond back and resolve your subscriber's questions is just amazing.
Sir my RMSEA value was .000
How can i justify or improve it???
Plz help
It could be due to variety of reasons. You may have a reasonable sample size with simple model.
It is already should perfect fit. This has to do more with justification than improvement.
@@researchwithfawad thank you sir... If you know any paper related to it plz let me know..
Type the following in Google Scholar to find relevant papers
allintext:"Perfect fit" "RMSEA=0.000" "AMOS"
or
allintext:"Perfect fit" "RMSEA = 0.000" "AMOS"
@@researchwithfawad thank you so much sir... God bless u sir 🙏🙏🙏
Assalamulaikum. can you share your data file to practice.
Please visit www.researchwithfawad.com
I need your help sir
For Queries, You can email me on kh.fawad83@gmail.com