12. SPSS AMOS - Assess Discriminant Validity - Fornell and Larcker Criterion

[@abebenegeri1823]

Thank you very much, Dr. Fawad. It is so helpful. Be blessed dear!!!!

[@researchwithfawad]

Thanks. Ameen. I am glad you liked it

[@mudasirmohammedibrahim1026]

Dear Dr Fawad. Please I am performing a confirmatory factor analysis on academic help seeking scale, which is made up of 3 subscales (instrumental, avoidant, and executive).
1. instrumental and avoidant are negatively correlated.
2. Instrumental and executive are negatively correlated.
3. Executive and avoidant are positively correlated.
During the factor analysis, I want to test for the Fornell and Larcker validity but I find it difficult to square root the negative correlations.
Now, my questions are;
1. Is it problematic when you have negative correlations between two or more constructs and/or subscales of a particular construct.
2. In the case of the above mentioned negative correlations, how do I go about it?
Sample size is 271, composite reliability for instrumental, executive, and avoidant are 0.87, 0.96, and 0.94 respectively.
Also, the AVE for instrumental, executive, and avoidant are 0.33, 0.59, and 0.52.
Model fit
CMIN/DF = 2.056
NFI= 0.831
RFI= 0.816
IFI= 0.905
TLI=0.896
CFI= 0.905
RMSEA= 0.063
SRMR= 0.0606
I have not performed modification indices to improve my model yet. But Ive observed that naturally the underlying contruct indicators do have negative correlations as reported in some literature.
Thank you

[@researchwithfawad]

You need to square root the AVE and then compare it to the correlation.

[@mudasirmohammedibrahim1026]

@@researchwithfawad Thank you sir

[@MeandYoutube516]

Dear Dr Fawad, thank you for your videos. what does it mean when some of the correlations are negative e.g. -0.754. Thanks

[@researchwithfawad]

Thanks for watching. This shows there is a negative relationship between the two variables.

[@ERFL1988]

Dear Dr. Fawad! Thank you for your amazing videos, very useful to familiarize ourselves with these topics.
I have a question, applying the Fornell & Larcker method, in your video, the square root of the AVE of each latent variable (LV) should be greater than its correlations with any other LV in the assessment.
However, I have seen that other researchers do the opposite, i.e. the square root of the correlations and compare it with the AVE of each LV. My questions are: Are both processes equivalent? Which method do I choose?
Thanks again for your amazing videos. All the best to you.

[@researchwithfawad]

Thanks. I am glad you liked it.
You can do both. Compare AVE with square of correlation or Square Root of AVE with the correlation.

[@ERFL1988]

@@researchwithfawad Thank you so much!

[@thechenmin]

@thechenmin
0秒钟前
Thank you so much prof. It is really helpful.
is the fornell-larcker ratio same as fornell alcker criterien???

[@researchwithfawad]

Yes, correct

[@tahmasbt8951]

Dear Fawad, thanks for your fabulous videos. My model includes an exogenous (predictor) variable, while yours doesn't. Thus, when I run the program to analyze discriminant validity and to see the correlation between variables, AMOS does not include the predictor variable in its report. Can you advise me on how to solve the issue?

[@researchwithfawad]

Thanks for watching. AUT is predictor and LS is dependent. The model results should show. Any errors. Is the model running fine.

[@KarthickSreVasan30]

Sir, do you upload the rectification video? As you mention at the last if yes pls highlight it sir.... I too have same issue of cross loading

[@researchwithfawad]

Please watch
ruclips.net/video/cFwczuJPmss/видео.html

[@jovial0731]

Thank you so much prof, I've watched almost all of your videos regarding SPSS AMOS - CFA and SEM. Super helpful for a newbie like me!!🥲
I just have a question, can I use HTMT ratio instead of F&L to calculate discriminant validity for CFA? In my case, discriminant validity is not valid with F&L criterion but HTMT, thank you in advance!!!

[@researchwithfawad]

Pleasure. I am glad you liked it. Sure you can.

[@jovial0731]

@@researchwithfawad I recommended your channel to my classmate also 🥰 Thank you for your reply!!

[@shadeel9293]

Dear, thank you for such an informative video. For 7 variables, we will also use the same procedure to investigate the discriminant Validity?

[@researchwithfawad]

Thanks. I am glad you liked it. The process is same for any number of variables

[@shadeel9293]

​@@researchwithfawadsir which version of Amos are you using in these videos?

[@saneshkp9810]

Sir, are the fit indices of the CFA and SEM models is same ?

[@researchwithfawad]

Yes

[@Alhamzah_F_Abbas]

Interesting. prof. if we have sub-dimensions is it better go with PLS instead of using AMOS? as I saw even though we trying to solve the AVE issue it is still there.

[@researchwithfawad]

Both are fine. Depending on the data as well.

[@jannatulmubasshera5535]

Sir, is there any reference of merging two construct as you mentioned in the last part?
and do we need to mention this in thesis paper?

[@researchwithfawad]

If they measure same construct and are in line with your conceptualization, you can merge.

[@DrNishayadav]

sir, are these excel add-on packages? watched the entire series in one go. sir your videos are too good.

[@researchwithfawad]

Thanks. I am glad you found the videos helpful. Please do circulate in your research circle.
No, excel is not addon. There are basic sheets i did myself.

[@talhamansoor7108]

Thank you

[@researchwithfawad]

Pleasure