Application of Structural Equation Models to Estimate Reliability and Validity of Psychological Tests and Assess the Reliability and Validity of Adult ADHD Diagnostic Scale (CAARS-S: SV)

Two important objectives for measuring the variables in behavioral and cognitive sciences are: 1) measurement and 2) complex causal relationships between variables. Measurement asks about, what is measured by observed variables? And how we can determine the validity and reliability of variables? The latter, responds to this question: since the variables are not directly observable and measurable and only reflected in observed variables with measurement error, how complex causal relationships between these variables can be inferred? How the relationships between latent variables can be measured? In response to these kinds of questions, structural equation models, composed of the "measurement model" and the "structural function model", are used. The answers to the first question are evaluated by the measurement model and that of the second one are evaluated by the structural function model. The structural equation models, emerging two models of Confirmatory Factor Analysis and Structural Function Analysis, eradicates many of difficulties in measuring latent variables and inference of the complex casual relationships among them. The present paper introduced the assumptions of structural equation models, presenting an example in the field of psychological test validation. It also explained the logic and structure of structural equations. CAARS–S: SV, a set of criteria for ADHD identification in adults, is used for explaining the confirmatory factor analysis