HTMT Discriminant validity

Small talk on SmartPLS that does not correspond to the other forums!
Post Reply
PLS Junior User
Posts: 8
Joined: Fri Aug 23, 2019 11:32 pm
Real name and title: Ana Foster, PhD student

HTMT Discriminant validity

Post by MaFa » Sat Nov 09, 2019 3:38 pm

Hello colleagues,

In my model, I have three constructs that are very related to each other, so both the correlations are above 0.7 and also HTMT values are 0.9 for them! I do not know what I can do, loadings are ok though. I have removed some items, but still, the issue is there. The CI values for HTMT in bootstrapping is lower than 1 though. Any suggestions would be highly appreciated.

SmartPLS Developer
Posts: 1110
Joined: Tue Mar 28, 2006 11:09 am
Real name and title: Dr. Jan-Michael Becker

Re: HTMT Discriminant validity

Post by jmbecker » Wed Dec 04, 2019 8:54 am

Given the statistics, discriminant validity at least may be a problem. You need to deal with that. Either by arguing why you think that the constructs are conceptually different despite the opposite finding from the statistics (and why the discriminant validity may not be a problem in your model), or by making changes to your model (e.g., you may model them as a higher-order component or something like this).
Maybe it is also evidence against your theory. Or you need to recollect data and make sure to measure the construct differently.

There are actually no clear and good guidelines in the literature on what to do when discriminant validity fails. It requires some creativity and scientific judgment on your side as you know the problem domain best.
Dr. Jan-Michael Becker, University of Cologne, SmartPLS Developer
GoogleScholar: ... AAAJ&hl=de

PLS Junior User
Posts: 1
Joined: Wed Dec 11, 2019 5:42 pm
Real name and title: Smith Clarkson

Re: HTMT Discriminant validity

Post by smithclarkson001 » Wed Dec 11, 2019 7:48 pm

The heterotrait-monotrait ratio of correlations (HTMT) is a new method for assessing discriminant validity in partial least squares structural equation modeling, which is one of the key building blocks of model evaluation. If discriminant validity is not established, researchers cannot be certain that the results confirming hypothesized structural paths are real, or whether they are merely the result of statistical discrepancies. The HTMT criterion clearly outperforms classic approaches to discriminant validity assessment such as Fornell-Larcker criterion and (partial) cross-loadings, which are largely unable to detect a lack of discriminant validity. VidMate download

Post Reply