Stages of health behavior change and mindsets: a latent class approach

Stage theories of health behavior are popular and of high practical relevance. Tests of the validity of these theories provide limited evidence because of validity and reliability problems. This study provides a bottom-up approach to identify behavioral stages from examining differences in underlying mindsets. We examine the concurrent validity of a latent-class-based approach and a commonly used stage-algorithm based on self-reports about intentions and behavior in order to identify possible strengths and shortcomings of previously used approaches.

Social-cognitive variables and individuals' stages were assessed in a sample of 2,219 internet users. Latent class analysis (LCA) was used to identify distinct groups with similar patterns of social-cognitive predictors. Convergent validity of the LCA solution and stage algorithms was tested by examining adjusted standardized residuals.

The LCA identified four distinct profiles-not intending to change, intending to change (no action), intending to change with action, and maintaining. Convergent validity with a stage algorithm was low, in particular in the nonintending and maintaining stages.

Stages as assigned by the stage-algorithm did not correspond well with the extracted mindsets: This indicates that commonly used stage-algorithms might not be effective in assigning individuals to stages that represent mindsets, undermining the possibility for stage-matched interventions.