Convergent and Discriminant Validity using SMART-PLS

Construct Validity

Validity is the assessment of whether a scale measures the concept of what it is intended to measure. Construct validity is assessed by establishing Convergent and Discriminant Validity. Both Convergent and Discriminant validity is established in reflectively measured constructs.

Convergent Validity: Convergent validity is established when items in a particular measure converge to represent the underlying construct. The AVE is calculated as the mean of the squared loadings of each indicator associated with a construct. Statistically, convergent validity is established when the Average Variance Extracted (AVE) is >0.50.

Discriminant Validity: Discriminant validity is established to ascertain the distinctiveness of the constructs in the study. It shows that constructs in the study have their own individual identity and are not too highly correlated with other constructs in the study. Discriminant validity in SMART-PLS is established using three different techniques.

    1. Fornell and Larcker Criterion: According to Fornell and Larcker Criterion, Discriminant validity is established if the Sq. The root of AVE for a particular construct is greater than its correlation with all other constructs.
    2. Cross Loadings: According to Cross loadings, a particular item should have higher loadings on its own parent construct in comparison to other constructs in the study. If an item loads well onto another construct in comparison to its own parent construct, then there are issues of discriminant validity. The difference of loading less than .10 also indicates that the item is cross loading onto the other construct and hence could be a threat to discriminant validity.
    3. Heterotrait-Monotrait (HTMT) Ratio: Based on prior research and their simulation study results, Henseler et al. (2015) suggest a threshold value of 0.90 if the path model includes constructs that are conceptually very similar (e.g., affective satisfaction, cognitive satisfaction, and loyalty); that is, an HTMT value above 0.90 depicts a lack of discriminant validity. However, when the constructs in the path model are conceptually more distinct, researchers should consider 0.85 as threshold for HTMT (Henseler et al. 2015).


Henseler, J., Ringle, C. M., & 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.

Sarstedt, M., Ringle, C. M., & Hair, J. F. (2021). Partial Least Squares Structural Equation Modeling. Springer.