Explaining General and Specific Factors in Longitudinal, Multimethod, and Bifactor Models: Some Caveats and Recommendations

Abstract An increasing number of psychological studies are devoted to the analysis of g-factor structures. One key purpose of applying g-factor models is to identify predictors or potential causes of the general and specific effects. Typically, researchers relate predictor variables directly to the general and specific factors using a classical mimic approach. However, this procedure bears some methodological challenges, which often lead to model misspecification and biased parameter estimates. We propose 2 possible modeling strategies to circumvent these problems: the multiconstruct bifactor and the residual approach. We illustrate both modeling approaches for the application of g-factor models to longitudinal and multitrait-multimethod data. Practical guidelines are provided for choosing an appropriate method in empirical applications, and the implications of this investigation for multimethod and longitudinal research are discussed. Translational Abstract Since the early years of personality research, many psychologists believe that psychological attributes involve a general (e.g., general intelligence) as well as several domain-specific (e.g., visual-spatial ability) components. A prominent statistical model for the assessment of general and specific components is the g-factor model (Spearman, 1914). Today, g-factor models are frequently applied in psychology and have been repeatedly recommended for the analysis of single-level data, multilevel data, longitudinal data, and multimethod data. The key purpose of applying g-factor models in practice is to identify determinants or potential causes of the general and specific components. For example, researchers may want to relate external variables (e.g., daily uplifts, daily hassles, or personality traits) simultaneously to general and specific factors (e.g., of intelligence or well-being). In this study, we show that relating external variables directly to the general and the specific factors leads to serious methodological problems. First, the model will no longer be the model that the researchers originally intended to fit to the data, meaning that the model will be misspecified. Second, the substantive conclusion may be false, because the parameters in the model may be biased. To circumvent these methodological problems, we propose two easy-to-apply methods that transform the explanatory variables in such a way that they can be safely linked to the general and specific factors in the models. Both trans