HTMT2–an improved criterion for assessing discriminant validity in structural equation modeling

Purpose One popular method to assess discriminant validity in structural equation modeling is the heterotrait-monotrait ratio of correlations (HTMT). However, HTMT assumes tau-equivalent measurement models, which are unlikely hold for most empirical studies. To relax this assumption, authors modify original and introduce a new consistent measure congeneric models: HTMT2. Design/methodology/approach The HTMT2 designed analogy but relies on geometric mean instead arithmetic mean. A Monte Carlo simulation compares performance In simulation, several design factors varied such as loading patterns, sample sizes inter-construct order compare estimation bias two criteria. Findings provides less biased estimations among latent variables compared HTMT, particular if indicators patterns heterogeneous. Consequently, should be preferred over case models. Research limitations/implications can only determined all between involved observable positive. Originality/value This paper introduces an improved version traditional HTMT. Compared other approaches assessing validity, advantages: (1) ease its computation, since based indicator correlations, (2) relaxed assumption tau-equivalence. highly recommend criterion