Entrepreneurial Cognition and Social Innovation: A Multilevel Analysis of Individual and Institutional Determinants

Background and Purpose: Social innovation (SI) has emerged as a critical mechanism for addressing a variety of social problems, especially in contexts where formal welfare systems are constrained. Social work research has recognized the growing role of entrepreneurial actors in shaping access to resources, services, and community well-being. However, limited empirical work has examined how individual cognitive orientations interact with broader institutional environments to produce socially innovative outcomes. This study draws on a cross-national dataset, to examine association between SI, entrepreneurial cognition and knowledge-transfer institutions.

Methods: This study uses data from the 2021 GEM Adult Population Survey (N = 8,415) across 28 countries. A multilevel modeling approach was employed to account for individuals nested within countries. The dependent variable is a composite measure of SI which captures SDG orientation, innovation novelty, and market reach. The key independent variable, cognitive orientation, reflects an entrepreneur’s vision and creativity. Country-level moderators include R&D transfer and post-school entrepreneurial education which represents institutional capacities for knowledge diffusion. Models were estimated using weighted regression with cluster-robust standard errors.

Results: We found that cognitive orientation is a robust and statistically significant predictor of social innovation across all model specifications (Model 3 (most stable): β = 0.115, SE = 0.017, p < .001). At the country level, post-school entrepreneurial education demonstrates a significant positive effect (β = 0.070, SE = 0.029, p = .025), while R&D transfer shows a marginal effect (β = 0.055, SE = 0.029, p = .073). Model fit improves modestly with the inclusion of institutional variables (R² increases from .042 to .072), indicating additional explanatory power. However, cross-level interaction effects are not statistically significant (COG × R&D: β = 0.019, SE = 0.020, p = .36; COG × post-school education: β = −0.005, SE = 0.021, p = .81), and contribute negligible variance (ΔR² = .001). Sub-dimensional analyses indicate that cognitive orientation strongly predicts SDG orientation (β = 0.250, p < .001) and innovation (β = 0.106, p < .001).

Discussion and Conclusion: These findings carry important implications for social work research and practice. First, they underscore the importance of individual-level cognitive capacities, such as vision and creativity, as drivers of socially meaningful innovation, aligning with strengths-based and empowerment-oriented approaches in social work. Second, institutional conditions act as significant predictors suggests that systemic investments in education and knowledge transfer can enhance social innovation at the population level. Third, the absence of interaction effects points to a critical insight: individual agency and institutional context operate as complementary, rather than conditional, forces. For social work policy, this implies that a push for social innovation should be complemented with investments in both human capabilities and institutional strengthening, rather than reliance on one domain alone. This study advances a multilevel understanding of social innovation that bridges micro-level cognition with macro-level systems and offers framework for promoting equitable and sustainable social change.