Discriminant validity

from Wikipedia, the free encyclopedia

Discriminant ( engl. Discriminant validity ), also discriminant validity , referred to in the multivariate statistical one aspect of the construct and is present when the measurements of various constructs differ. The concept of discriminant validity was introduced by Campbell and Fiske (1959).

Finding

Discriminant validity must be determined at both the construct and indicator level. At the construct level, the confirmatory factor analysis (CFA) and the multitrait multimethod approach are part of the standard methodological repertoire. In the latter case, the convergence validity and the discriminant validity are compared using a single sample. In short, it is expected that the convergence validity is greater than the discriminant validity.

Another common method at the construct level is the application of the Fornell-Larcker criterion as the result of an AVE-SV comparison: If the average recorded variance (AVE) of a construct is higher than any squared correlation with another construct (SV), then Discriminant validity can be assumed at construct level. (It should be noted here that the error-corrected correlations between constructs from the CFA model are used instead of the correlations taken from the raw data.) However, this quality measure has proven to be less reliable in simulation models for variance-based structural equation models (e.g. PLS), on the other hand, with covariance-based structural equation models (e.g. Amos) at the construct level it is very reliable.

A more recent method at the construct level has been published by Henseler et al. (2015) and is known as the heterotrait-monotrait ratio (HTMT). It provides reliable results for both variance-based and covariance-based structural equation models. Voorhees et al. (2015) suggest a combination of the AVE-SV comparison and the HTMT method for the latter, with the maximum value 0.85 being proposed as the HTMT rate. Discriminant validity at the indicator level can be determined, for example, by EFA .

Discriminant validity is only one building block for determining the construct validity of a construct. Further building blocks are convergence validity , nomological validity and content validity based on a definition of the construct. In addition to validity, reliability is also important; In one-dimensional measurement models it is usually determined as tau-equivalent reliability (traditionally also known as Cronbach's ) or congeneric reliability (traditionally also known as composite reliability ).

criticism

Discriminant and convergence validity are widely used as building blocks of construct validity . However, their consideration is mainly criticized by John R. Rossiter , who argues that the construct validity must be achieved independently of other constructs. He emphasizes the importance of content validity and equates it with construct validity. Measures to improve discriminant and convergence validity can lead to indicators being removed and the statistically measurable properties of the measurement models being improved, while at the same time the measurement models move away from the semantic content of their constructs. This view was in turn criticized by Adamantios Diamantopoulos , who points out the importance of discriminant validity, since this is the only way to ensure that two constructs actually measure something different.

swell

  1. ↑ a b D. T. Campbell, DW Fiske (1959): Convergent and discriminant validation by the multitrait-multimethod matrix. Psychological Bulletin, Volume 56, pp. 81-105, doi : 10.1037 / h0046016 .
  2. ↑ Bagozzi, Yi & Phillips (1991) Assessing construct validity in organizational research. Administrative Science Quarterly, 36, 421-458, JSTOR 2393203 .
  3. ^ Claes Fornell, David F. Larcker: Evaluating Structural Equation Models with Unobservable Variables and Measurement Error . In: Journal of Marketing Research . tape February 18 , 1981, p. 39-50 , JSTOR : 3151312 .
  4. ↑ a b J. Henseler, CM Ringle, M. Sarstedt, 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, doi : 10.1007 / s11747-014-0403-8 .
  5. ↑ a b c C.M. Voorhees, MK Brady, R. Calantone, E. Ramirez, 2016. Discriminant validity testing in marketing: an analysis, causes for concern, and proposed remedies. Journal of the Academy of Marketing Science, 44 (1), 119-134, doi : 10.1007 / s11747-015-0455-4 .
  6. ↑ John R. Rossiter (2008): Content Validity of Measures of Abstract Constructs in Management and Organizational Research. British Journal of Management, Volume 19, pp. 380-388, doi : 10.1111 / j.1467-8551.2008.00587.x .
  7. ^ A. Diamantopoulos (2005): The C-OAR-SE procedure for scale development in marketing: A comment. International Journal of Research in Marketing, Volume 22, pp. 1-9, doi : 10.1016 / j.ijresmar.2003.08.002 .