Explainable artificial intelligence for predicting entrepreneurial intentions: a theory of planned behavior approach

Purpose Several researchers have approached entrepreneurial intentions by studying personal attributes, demonstrating that individuals' attitudes and beliefs positively influence their entrepreneurial intentions. However, this area of research has been criticized for concentrating on behavioral intentions at the expense of cognitive components. In this regard, the theory of planned behavior (TPB) has become one of the most applied theoretical frameworks in this field of study. Nevertheless, less is understood about the prominence of the TPB dimensions, their predictive capacity, and the more complex non-linear relationships. To address the limitations of traditional regression methods in tackling these gaps, we utilize explainable artificial intelligence techniques to examine the dominance and nonlinearity of institutional dimensions in predicting entrepreneurial intention. Design/methodology/approach This study's data set is drawn from the 2020 Adult Population Survey (APS) conducted by the Global Entrepreneurship Monitor (GEM). The GEM APS is a comprehensive dataset presenting a representative sample of the adult population across various countries, including individual-level data related to entrepreneurial activities, intentions, and the factors influencing entrepreneurship (Bosma et al., 2021). The 2020 dataset includes responses from a total of 141,403 individuals across multiple countries, providing the basis for a thorough analyzis of entrepreneurship on a global scale. Findings The findings of this study highlight the complex relationships between the individual, contextual, and institutional factors that influence entrepreneurial intentions and behaviors as modeled through the TPB. The results also emphasize the significant role of individual factors, such as skills, knowledge and perceived opportunities, in shaping entrepreneurial activities. Variables such as self-efficacy, creativity and the perception of opportunities thus emerge as key predictors, aligning with the TPB dimensions of Attitude towards Behavior and Perceived Behavioral Control. Originality/value These findings enhance research on the TPB and entrepreneurial intention, emphasizing significant areas where machine-learning methods can advance entrepreneurship research and policy.