HTMT: A New Criterion to Assess Discriminant Validity

The heterotrait-monotrait ratio of correlations (HTMT) is a new criterion to assess discriminant validity in variance-based structural equation modeling, which is superior compared with the Fornell-Larcker criterion and (partial) cross-loadings; see
Hair, J. F., Hult, G. T. M., Ringle, C. M., and Sarstedt, M.(2022). A Primer on Partial Least Squares Structural Equation Modeling (3rd ed.). Thousand Oakes, CA: Sage. www.pls-sem.com
Henseler, J., Ringle, C. M., and Sarstedt, M. (2015). A New Criterion for Assessing Discriminant Validity in Variance-based Structural Equation Modeling. Journal of the Academy of Marketing Science, 43(1), 115-135. dx.doi.org/10.1...
The new HTMT criterion is implemented in SmartPLS: www.smartpls.com

[@taian3935]

Thanks very much. Could you please show us how to operate to test the discriminant validity in AMOS using HTMT inference?

[@faintersauce2813]

Can you please make a video describing factor analysis for both reflective and formative constructs

[@choukrimenidjel363]

Good morning, The authors, Henseler et al. (2015), P. 129, in the table 6, utilized the following expretion: CI.900 [0.612;0.652] I understand that CI means “Confidence Intervals”, but I do not understand what the two values [0.612;0.652] mean? Where do I find them in smartPLS Tank you

[@AntonioAlves-tf5vs]

Choukri MENIDJEL That's the interval. lower value: 0.612; higher value: 0.652

[@amsyardaud6628]

what the value 900 stand for?

[@spektrumspektrum6664]

Helo, how to remove some variabel which are not fit with HTMT criteria? I mean how to read that? As we know in SMART PLS result we have vertical column and horizontal column.

[@smartpls5336]

Just delete the variable from the PLS path model and compute the results again.

[@accountingfinancial8400]

hi dear thank u very very very much the best

[@carstenstobwasser2399]

In my model 3 of 120 relationships violate the HTMT Inference criterion. However, for all constructs the 0.85 and 0.90 cirterions are not violated at all (Also, there are no cross loadings and the Fornell-Larker test holds true). Does this mean I cannot establish discriminant validity?

[@smartpls5336]

Check our article; your results support that discriminant validity has been established.

[@faintersauce2813]

How can I generate factor analysis using Smartpls 3 I have three different fameworks and I'm using the same constructs to test the three models. I noticed when I generated the factor analysis for the whole model that the outweight are different between different models. I understood that factor analysis is to test the relation between the indicators and their coresponding construct and it should not vary between different models. Is that true? Please advise me

[@smartpls5336]

PLS-SEM is not a factor analysis but a context-specific model estimation. If you change the model (i.e., the context), the resulting latent variable scores change and thus the relationships with the indicators of each latent variable.

[@amsyardaud6628]

what are the value of 900 in the HTMT stand for? in the HTMT table

[@burstsoon]

What does HTMT mean in the layman terms? Or the simple meaning of HTMT? I had been attempted to search it but still can't comprehend it.

[@smartpls5336]

Heterotrait-monotrait Ratio of Correlations (HTMT): www.smartpls.com/documentation/algorithms-and-techniques/discriminant-validity-assessment

[@sarahoseingholizade9155]

Thanks for the video; Unfortunately I couldn't understand what to do when my Discriminant Validity is not established. What should I do? According to the paper, I must either Increase the monotrait-heteromethod correlations OR Decrease the heterotrait-heteromethod correlations. Well, how can I do that?

[@smartpls5336]

For example, you can reassign the indicators, combine the two constructs into one, or split one construct into two constructs, each with a more homogeneous set of indicators. However, all of this should be theoretically underpinned.

[@赵晓晓-i5b]

Dear Teacher: I have an HTMT value of 0.985 what should I do? Is there any way to debug the problem. Thank you.