Cited by 38,459Psychometric Methodologies and TestingStatistical Methods and ApplicationsBusiness Strategies and Management Research

A new criterion for assessing discriminant validity in variance-based structural equation modeling

Open access
Henseler, Jörg, Ringle, Christian M., Sarstedt, Marko·2015·Journal of the Academy of Marketing Science

Abstract

Discriminant validity assessment has become a generally accepted prerequisite for analyzing relationships between latent variables. For variance-based structural equation modeling, such as partial least squares, the Fornell-Larcker criterion and the examination of cross-loadings are the dominant approaches for evaluating discriminant validity. By means of a simulation study, we show that these approaches do not reliably detect the lack of discriminant validity in common research situations. We therefore propose an alternative approach, based on the multitrait-multimethod matrix, to assess discriminant validity: the heterotrait-monotrait ratio of correlations. We demonstrate its superior performance by means of a Monte Carlo simulation study, in which we compare the new approach to the Fornell-Larcker criterion and the assessment of (partial) cross-loadings. Finally, we provide guidelines on how to handle discriminant validity issues in variance-based structural equation modeling.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1007/s11747-014-0403-8

Cites

108 of 108 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Henseler et al. (2015) did not recommend HTMT inference because it failed to detect discriminant validity violations even when inter-construct correlations were as high as 0.95.

    “Ultimately, they did not recommend HTMT inference , given it failed to detect discriminant validity violations even when inter-construct correlations were as high as 0.95.”

    From Reporting reliability, convergent and discriminant validity with structural equation modeling: A review and best-practice recommendations (2024) · cites this paper as Henseler et al. (2015)

    The Specificity results report that HTMT inference often treats constructs as distinct even when their correlation reaches 0.95, supporting the stated concern about its liberality. But the Key findings and recommendations explicitly recommend HTMT inference as an option in some settings, contrary to the claim that the authors did not recommend it. Page numbers are unavailable in our copy. (LLM, conf=0.91)

    Checked by a reasoning model — may be wrong.

    Experimental — accuracy not yet measured
  • Henseler et al.'s (2015) HTMT equation is incorrect for estimating the disattenuated construct score correlation.

    “First, intending to estimate the disattenuated construct score correlation, Henseler et al.'s (2015, Eq. 6) HTMT equation is incorrect.”

    From Reporting reliability, convergent and discriminant validity with structural equation modeling: A review and best-practice recommendations (2024) · cites this paper as Henseler et al. (2015)

    The paper does not say its Eq. 6 is incorrect. In the HTMT discussion (page unavailable), it describes HTMT as an estimate of the inter-construct correlation that parallels the disattenuated construct score correlation; the Appendix (page unavailable) says it demonstrates that HTMT is an estimator of that correlation. The paper thus presents the equation as valid rather than identifying the error alleged in the citing sentence. (LLM, conf=0.86)

    Checked by a reasoning model — may be wrong.

    Experimental — accuracy not yet measured
  • Henseler et al.'s (2015) HTMT equation becomes the disattenuated correlation between two composite scores formed by averaging the standardized item scores.

    “While Henseler et al.'s (2015) HTMT equation uses inter-item correlations instead of inter-item covariances, that becomes the disattenuated correlation between two composite scores formed by averaging the standardized item scores.”

    From Reporting reliability, convergent and discriminant validity with structural equation modeling: A review and best-practice recommendations (2024) · cites this paper as Henseler et al. (2015)

    The empirical example and Appendix define HTMT using inter-item correlations, and the main discussion says it parallels a disattenuated construct-score correlation. The paper does not state that HTMT becomes the disattenuated correlation of two composites formed by averaging standardized item scores; it distinguishes HTMT from a method that requires calculating construct scores. (LLM, conf=0.85)

    Experimental — accuracy not yet measured
  • The recommended HTMT threshold of 0.85 was based on the power of detecting a population correlation at 1.0 in a simulation study.

    “Third, the recommended HTMT threshold of 0.85 was based on the power of detecting a population correlation at 1.0 in a simulation study (Henseler et al., 2015).”

    From Reporting reliability, convergent and discriminant validity with structural equation modeling: A review and best-practice recommendations (2024) · cites this paper as Henseler et al., 2015

    The Sensitivity results report that HTMT .85 detected a lack of discriminant validity in 99.90% of simulated cases, so the paper does assess its detection power. But the section introducing the threshold says 0.85 was suggested by earlier authors and treats it as a predefined, debatable cutoff; it does not say the simulation’s power results determined that value. Our copy provides no page number for these passages. (LLM, conf=0.86)

    Experimental — accuracy not yet measured
  • Unless the construct scores are calculated by averaging the standardized item scores, Henseler et al.'s HTMT equation provides a biased estimate of the disattenuated correlation between two composites.

    “Hence, unless the construct scores are calculated by averaging the standardized item scores, Henseler et al.'s HTMT equation provides a biased estimate of the disattenuated correlation between two composites.”

    From Reporting reliability, convergent and discriminant validity with structural equation modeling: A review and best-practice recommendations (2024) · cites this paper as Henseler et al. (2015)

    The HTMT discussion (section “b) a)”; page not provided) says HTMT estimates the correlation between constructs and parallels a disattenuated construct-score correlation. It also says HTMT does not require construct scores. The cited paper does not state that HTMT is biased for correlations between composites unless scores are calculated by averaging standardized items. (LLM, conf=0.85)

    Experimental — accuracy not yet measured

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

Flagged first, then by citations

4 of the 112 references couldn’t be matched to a known paper.

Research works printed but not linked: 2

  1. Gefen, D., Straub, D. W., & Boudreau, M.-C. (2000). Structural equation modeling techniques and regression: guidelines for research practice. Communications of the AIS, 4, 1–78.
  2. Ringle, C. M., Sarstedt, M., & Straub, D. W. (2012). A critical look at the use of PLS-SEM in MIS Quarterly. MIS Quarterly, 36(1), iii–xiv.

Cited by

Sorted by citations

Authors
Henseler, J., Ringle, C. M., & Sarstedt, M.
Affiliations
  • Universidade Nova de Lisboa (PT)
  • University of Newcastle Australia (AU)
Corresponding author
Sarstedt, Marko
Venue
Journal of the Academy of Marketing Science
Published
2014-08-22
Volume
43
Issue
1
Pages
115-135
ISSN
  • 0092-0703 (ISSN-L)
  • 1552-7824
OpenAlex
W2105846236
Cited by
38,459
References
112
Citations by year
2026: 9160 · 2025: 7076 · 2024: 5683 · 2023: 5082 · 2022: 4007 · 2021: 2951 · 2020: 1857 · 2019: 1210 · 2018: 761 · 2017: 384 · 2016: 208 · 2015: 58 · 2014: 7 · 2013: 1 · 2012: 2
FWCI
276.66
Type
journal article
Language
en
Open access
Locations
  • Journal of the Academy of Marketing Science (publishedVersion, cc-by)
  • New University of Lisbon's Repository (New University of Lisbon) (publishedVersion, other-oa)
  • University of Twente Research Information (acceptedVersion, cc-by)
  • Figshare (submittedVersion, cc-by)
  • Portuguese National Funding Agency for Science, Research and Technology (RCAAP Project by FCT) (submittedVersion, other-oa)
  • tub.dok (Hamburg University of Technology) (submittedVersion)
  • NOVA (University of Newcastle Australia) (submittedVersion, cc-by)
  • University of Twente Research Information (submittedVersion, cc-by)
Funders
  • Rijksuniversiteit Groningen
  • Radboud Universiteit
Topics
  • Social Sciences › Decision Sciences › Management Science and Operations Research › Psychometric Methodologies and Testing
  • Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods and Applications
  • Social Sciences › Business, Management and Accounting › Strategy and Management › Business Strategies and Management Research
Keywords
discriminant validity, partial least squares, cross-loadings, multitrait-multimethod matrix, latent variables, measurement validity

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review