Discriminant Validity Assessment: Use of Fornell & Larcker Criterion versus HTMT Criterion

Assessment of discriminant validity is a must in any research that involves latent variables for the prevention of multicollinearity issues. Fornell and Larcker criterion is the most widely used method for this purpose. However, a new method has emerged for establishing the discriminant validity ass...

Full description

Bibliographic Details
Main Authors: M. R., Abdul Hamid, Sami, W., M. H., Mohmad Sidek
Format: Conference or Workshop Item
Language:English
Published: IOP Publishing Ltd 2017
Subjects:
Online Access:http://umpir.ump.edu.my/id/eprint/20095/
http://umpir.ump.edu.my/id/eprint/20095/
http://umpir.ump.edu.my/id/eprint/20095/1/Ab%20Hamid%20et%20al.%202017_J._Phys.__Conf._Ser._890_012163.pdf
Description
Summary:Assessment of discriminant validity is a must in any research that involves latent variables for the prevention of multicollinearity issues. Fornell and Larcker criterion is the most widely used method for this purpose. However, a new method has emerged for establishing the discriminant validity assessment through heterotrait-monotrait (HTMT) ratio of correlations method. Therefore, this article presents the results of discriminant validity assessment using these methods. Data from previous study was used that involved 429 respondents for empirical validation of value-based excellence model in higher education institutions (HEI) in Malaysia. From the analysis, the convergent, divergent and discriminant validity were established and admissible using Fornell and Larcker criterion. However, the discriminant validity is an issue when employing the HTMT criterion. This shows that the latent variables under study faced the issue of multicollinearity and should be looked into for further details. This also implied that the HTMT criterion is a stringent measure that could detect the possible indiscriminant among the latent variables. In conclusion, the instrument which consisted of six latent variables was still lacking in terms of discriminant validity and should be explored further.