Guidelines for Construct Measurement Yielding Unexpected Higher-Order Constructs: An Application for the Theory of Planned Behavior Applied to Condom Use

The theory of planned behavior (TPB) is useful theory in social science for explaining how people arrive at an intention to act. Previously, distinctions were found among the global theoretical constructs (attitudes, subjective norms, and volition). As such, six higher-order constructs, or differentiated constructs , were identified. The research investigation intended to validate the use of the TPB differentiated constructs for construct measurement, which is useful for latent variable models. Results based on studies using latent variable methodologies are at fault for improper measurement specification which may call into question the implications of the research. Using Bollen and Lennox’s conventional guidelines for construct measurement, survey data obtained from 446 African American adolescents about their condom use beliefs was used to explore the applicability of the first conventional guideline for construct measurement: establishing linear combinations of construct items with the use of principal axis factoring. Secondary analyses examined the four remaining guidelines. The six hypothesized constructs identified were: affective and instrumental attitude, descriptive and injunctive norms, perceived controllability, and self-efficacy. Surprisingly, results indicated three additional constructs, not previously explained by the theory: negative affective attitude, corollary condom use attitude, and partnership self-efficacy. These findings highlight not only the need to apply such guidelines to all research analyses involving construct measurement where appropriate, but also illustrate how unexpected findings through analytical best practices can expand and modify existing theoretical conceptualizations.