A theoretical and philosophical revision of the concept of fit in structural equation modeling and its relation to a confirmation bias is developed. The neutral character of fit indexes regarding this issue is argued, concluding that protection against confirmation bias relies on model modification strategy and scientist behavior. (SLD)

Provides an introduction to the structural equation modeling concepts developed by J. Pearl, discussing the concept he calls "d-separation." Explains how d-separation connects to control variables, partial correlations, causal structuring, and even a potential mistake in regression. (SLD)

Used three approaches to understand the effect of the number of variables in the model on model fit in structural equation modeling through computer simulation. Developed a simple formula for the theoretical value of the comparative fit index. (SLD)

Reviews the use of structural equation modeling for providing an overall assessment of mediation, the mechanism that accounts for the relation between the predictor and the criterion. A strategy for supplemental details is presented that measures the magnitude of mediational effects. (SLD)

Problems found in the application of structural equation modeling to the theory of reasoned action are explored, and an alternative model specification is proposed that improves the fit of the data while leaving intact the structural part of the model being tested. Problems and the proposed alternative are illustrated. (SLD)

A Monte Carlo study evaluated the effectiveness of different factor analysis extraction and rotation methods for identifying the known population multiple-indicator measurement model. Results demonstrate that exploratory factor analysis can contribute to a useful heuristic strategy for model specification prior to cross-validation with…

Proposes a novel structural modeling approach based on latent growth curve model specifications for use with dyadic data. The approach allows researchers to test more sophisticated causal models, incorporate latent variables, and estimate more complex error structures than is currently possible using hierarchical linear modeling or multilevel…

Proposes an analog to the Scheffe test (H. Scheffe, 1953) to be applied to the exploratory model-modification scenario. The method is a sequential finite-intersection multiple-comparison procedure that controls the Type I error rate to a desired alpha level across all possible post hoc model modifications. (SLD)

Introduces an alternative structural equation modeling (SEM) specification search approach based on the Tabu search procedure. Using data with known structure, the procedure is illustrated, and its capabilities for specification searches in SEM are demonstrated. (Author/SLD)

Provides counterexamples where the covariance matrix provides crucial information about consequential model misspecifications and cautions researchers about overinterpreting the conclusion of D. Rogosa and J. Willett (1985) that the covariance matrix is a severe summary of longitudinal data that may discard crucial information about growth. (SLD)

Illustrates how structural equation modeling can be used to test the multidimensionality of a measure. Using data collected on a multidimensional measure, compares an oblique factor model with a higher order factor model, and shows how the oblique factor model fits the data better. (SLD)

How is affective change rated with positive adjectives such as good related to change rated with negative adjectives such as bad? Two nested perfect and imperfect forms of dynamic bipolarity are defined using latent change structural equation models based on tetrads of items. Perfect bipolarity means that latent change scores correlate -1.…

In previous research (Hu & Bentler, 1998, 1999), 2 conclusions were drawn: standardized root mean squared residual (SRMR) was the most sensitive to misspecified factor covariances, and a group of other fit indexes were most sensitive to misspecified factor loadings. Based on these findings, a 2-index strategy-that is, SRMR coupled with another…

A didactic example is presented of the application of new developments in structural equation modeling that allow for the modeling of multilevel data. The method, a synthesis of methods developed by B. Muthen, is applied to the problem of validating indicators of science education quality in the United States. (SLD)

Makes the case that the Raykov and Marcoulides Proof (RMP; T. Raykov and G. Marcoulides, 2001) generalizes to a broad class of structural equation (SE) models. Suggests a counterexample that accepts the statistical dimensions of the RMP while questioning the conclusion, highlighting the need for greater attention to the semantic dimension of SE…