Predicting crowdfunding success through gender dynamics and social validation: a comparative analysis using econometrics and machine learning models

Purpose This study aims to examines the influence of gender and social validation – proxied by the number of campaign supporters – on the success of reward-based crowdfunding campaigns. It explores whether female entrepreneurs experience disparities in funding outcomes and assesses the extent to which social endorsement mitigates or amplifies these effects. Design/methodology/approach This study uses a dual-method approach, combining traditional logistic regression with supervised machine learning (ML) models to examine the effects of gender and supporter count on crowdfunding success (1 = successful, 0 = unsuccessful). Using data from 561 campaigns (188 female-led, 373 male-led), eight ML models – Bagging, Decision Tree, Logistic Regression, Random Forest, SVM, XGBoost, Deep Learning (MLP) and Adaptive Tree Induction (ATI) – were trained and evaluated using accuracy-based metrics. This comparative analysis assesses the relative effectiveness of logistic regression versus ML models in predicting crowdfunding success. Findings Supporter count consistently emerges as the strongest associational predictor of crowdfunding success, with early social backing particularly benefiting female-led projects. Gender alone does not drive crowdfunding success, but its interaction with supporters highlights the role of social validation in overcoming credibility barriers. Across classification models, XGBoost, SVM and Bagging Classifier achieve top predictive performance, while logistic regression remains a reliable, interpretable option. Feature importance and model rankings consistently emphasise community support, supporter–gender interactions, social engagement and project visibility as key predictive features associated with crowdfunding success. Research limitations/implications While the analysis is limited to a single country and platform, potentially constraining generalisability, the findings offer valuable insights. Future research should explore cross-platform and cross-country comparisons to enhance external validity. Despite these limitations, the results have practical implications for female entrepreneurs, crowdfunding platforms and policymakers seeking to address gender bias and foster inclusive access to entrepreneurial finance. Originality/value To the best of the authors’ knowledge, this study is among the first to use predictive analytics using ML models to investigate the determinants of crowdfunding success and if there exist gender disparities in crowdfunding within the Malaysian context. By assessing both the determinants and predictability of campaign success, it offers a nuanced understanding of the gender gap – examining not only why certain campaigns succeed but also how effectively success can be predicted based on campaign characteristics.