Business idea possession as the robust predictor of entrepreneurial intention: a triangulated baseline among Ecuadorian business students

Although sociodemographic characteristics feature prominently in entrepreneurial intention research, the magnitude of their direct associations—estimated without psychological mediators—remains insufficiently documented. This study examines these associations among 500 business administration students from six Ecuadorian universities, using structural regression models with a latent intention outcome (MIMIC-type specifications), triangulated across three analytical specifications. Business idea possession emerged as the sole robust predictor across all specifications; age reached significance only after data cleaning, and entrepreneurial parents showed at most a marginal trend. Gender, education, income, access to finance, and training exposure showed null associations, confirmed as negligible for gender and training hours through equivalence testing. Sociodemographic predictors explained under 15% of intention variance. The intention construct showed scalar measurement invariance across gender, with no structural or mean differences. These findings establish a descriptive baseline: modest explained variance is consistent with—but does not test—mediation frameworks, and equally consistent with null direct effects or model underspecification. Rather than demonstrating that sociodemographic characteristics matter—an already-established premise—the study’s contribution is a methodological triangulation combined with formal invariance and equivalence testing that separates robust from fragile associations, yielding a verified benchmark for future mediation research. Implications for entrepreneurship education emphasize facilitating concrete business idea generation over uniform content exposure.