Unobservable Assumptions: Is “Self-Control” a Reflective Measure, Formative Construct, or Measurement Fiction?

Criminologists typically invoke theoretical constructs within an implicit reflective measurement paradigm, assuming constructs represent unified entities causing observable indicators. Yet measurement model choices embed fundamental theoretical commitments rarely subjected to systematic examination. Using self-control as a paradigmatic example, we analyze an abbreviated 12-item Grasmick et al. (1993) scale across 30 countries in the ISRD-2 data through confirmatory factor analysis, network analysis, and predictive concordance assessment. Results challenge unidimensional reflective assumptions: modest model fit masks substantial item-level heterogeneity; systematic cross-cultural measurement non-invariance emerges; network structures suggest complex interdependencies rather than common-cause relationships; and substantial individual-level prediction discordance reveals a “correlation paradox” where aggregate statistical success corresponds to individual-level theoretical failure. These patterns suggest alternative structures—multidimensional reflective, formative, or fictional—merit consideration. Persistent anomalies may reflect theorists struggling to articulate complex conceptualizations within disciplines lacking appropriate measurement vocabularies. Rather than advocating specific approaches, we argue for explicit measurement choices and appropriate validation methods. The disconnect between aggregate success and individual-level validity may characterize research programs trapped in degenerating development patterns.