Construct validation of questionnaires and tests is often done, these days, by means of confirmatory factor analysis with respect to the predicted factor structure (which should follow the various test scales). However, one disadvantage of CFA is that secondary factor loadings are not part of the output. Therefore, it is not possible to determine whether a certain item would perhaps have been better assigned to another cluster, especially when its primary factor loading is low. A consequence is perhaps that the deviation from the prediction becomes less visible; moreover, there are less starting points for suggesting a possible revision of the prediction.

That may be the reason that, in a number of cases, the investigators content themselves with little more than reporting a few GOF indices for the "winning" and a few rival models, concluding that the model is acceptable or confirmed. They may do so, even if the values of some of the fit indices are mediocre, some of the factor indicator loadings are clearly too low, and some modification indices indicate the necessity of modification of the model. Notably, in the case of a newly devised instrument, being psychometrically investigated for the first time, this should not be the path to follow. The sophistication of SEM programs, in combination with impressive-looking but ill-understood fit indices, seems to be an excuse to refrain from critically investigating sources and the meaning of (remaining) misfit.

In addition, how reliable are these indices of goodness of fit in the first place? There has been much critical discussion about GOF indices in the specialized literature and on a special discussion website for SEM users (SEMNET). From this discussion, one can learn that, for a model test, 2 and the GOF indices are unreliable, and for an estimation of degree of fit, the GOF indices are too imprecise.

The difficulties in applying and interpreting GOF indices made Barrett (2007) call for their abandonment: “I would recommend banning ALL such indices from ever appearing in any paper as indicative of ‘model acceptability’ or ‘degree of misfit’." (p. 821). I agree with this point of view.

For an extensive discussion of this problem, see my publication on Researchgate: Confirmatory factor analysis: a brief introduction and critique.

Barrett, P. (2007). Structural equation modeling: adjudging model fit, Personality and Individual differences, 42 (5), 815–824.

More Peter Prudon's questions See All
Similar questions and discussions