Psychometric Network Analysis of the Hungarian WAIS

Abstract

The positive manifold—the finding that cognitive ability measures demonstrate positive correlations with one another—has led to models of intelligence that include a general cognitive ability or general intelligence (g). This view has been reinforced using factor analysis and reflective, higher-order latent variable models. However, a new theory of intelligence, Process Overlap Theory (POT), posits that g is not a psychological attribute but an index of cognitive abilities that results from an interconnected network of cognitive processes. These competing theories of intelligence are compared using two different statistical modeling techniques: (a) latent variable modeling and (b) psychometric network analysis. Network models display partial correlations between pairs of observed variables that demonstrate direct relationships among observations. Secondary data analysis was conducted using the Hungarian Wechsler Adult Intelligence Scale Fourth Edition (H-WAIS-IV). The underlying structure of the H-WAIS-IV was first assessed using confirmatory factor analysis assuming a reflective, higher-order model and then reanalyzed using psychometric network analysis. The compatibility (or lack thereof) of these theoretical accounts of intelligence with the data are discussed.

The Effects of Using Partial or Uncorrected Correlation Matrices When Comparing Network and Latent Variable Models

“…These methods account for the covariance between test scores with latent variables. Network models have been proposed as solutions to the problem of modeling psychological constructs such as the symptoms of psychopathology (Cramer et al 2010) and more recently, cognitive test scores (Kan et al 2019;Schmank et al 2019). The network models proposed by these investigators do not make use of latent variables, but rather model the structure of test batteries in terms of manifest variables (i.e., observed test scores).…”

“…Psychometric network models represent correlation matrices as a graph in which each variable is a node and each correlation an edge. Both Kan et al (2019) and Schmank et al (2019) used reduced partial correlation matrices of WAIS-IV test scores. The rational for the use of partial correlations is that one can rule out the relationship between any pair of variables as being due to other variables in the analysis and thus more readily infer causation .…”

“…Both Kan et al (2019) and Schmank et al (2019) compared model fit indices for network models with those for traditional latent variable models and found the former to provide a better fit. Kan et al (2019) suggested that their results provided support for a mutualism model (van der Maas et al 2006).…”

“…According to Kan et al (2019), mutualism assumes that reciprocal causation occurs among cognitive abilities during development. The positive manifold (i.e., the positive correlation between cognitive test scores) is due to these interactions rather than g. Schmank et al (2019) suggested that network models are compatible with process overlap theory (Kovacs and Conway 2016).…”

“…From the description of the methods by Kan et al (2019) and Schmank et al (2019), it is not clear what type of correlations were used to compare network and latent variable models. However, from the R code provided by Schmank et al (2019), it is clear that the fit of network models to partial correlations were compared to the fit of latent variable models to uncorrected correlation matrices.…”