Veranstaltungsprogramm der VHB Jahrestagung

Predicting peer donor transformation using machine learning

Peer-to-peer fundraising has emerged as a popular funding approach for nonprofit organizations, generating fast revenue and a promising opportunity for donor base expansion by transforming peer donors into organizational donors following their peer donation (Hesse & Boenigk, 2023). The paper discusses the “transformation likelihood,” which is the probability of peer donors directly contributing to the nonprofit organization beyond their initial peer donation. Thereby, the aim of the paper is twofold. First, it aims to identify determining factors of the transformation likelihood. Second, it evaluates if using artificial intelligence (machine learning) can assist in identifying peer donors most likely to undergo transformation.

The data was collected in July 2023 through an online survey targeting U.S. donors that have contributed to a peer-to-peer fundraising campaign before, resulting in a final sample of N =706. The study’s data analysis is twofold. To test the factors affecting the transformation likelihood, and considering the binary nature of the outcome variable, logistic regression is performed. Second, to evaluate the effectiveness of employing machine learning to identify which of the previously mentioned factors accurately predicts transformation likelihood at a donor-level, supervised learning is applied, which is commonly used for predictive tasks by leveraging existing labeled data to predict future events (Casado et al., 2023).

Among peer donors unaffiliated with the nonprofit before, the transformation likelihood is 14.1%. Utilizing a random forest classifier, the study achieves a 77% accuracy in predicting transformation, identifying key factors for accurate prediction. Interestingly, the study challenges some preconceived notions about transformation behavior in traditional fundraising contexts

Non-profits have professionalized over the last decades, which makes them increasingly resemble for-profits (Hwang & Powell, 2009; Maier et al., 2016). For-profits, on the other hand, have adopted the language of non-profits, as the heightened importance of corporate social responsibility exemplifies (Plummer et al., 2020). Thus, as has been thoroughly documented, the boundaries between two sectors that were once conceptualized as following distinct institutional logics have become blurry (Bromley & Meyer, 2017; De Bakker et al., 2013). To investigate the ways in which the two sectors became similar over time, we make use of a longitudinal dataset covering all registered Swiss for-profits (~1.000.000) and most Swiss non-profits (~60.000) between 2003 and 2022. Information on the organizations was obtained through the publicly available Swiss Registry of Commerce. Most importantly, the registry contains the mission statements of all organizations, which allows us to compute the similarity of stated missions between for-profits and non-profits. Our results show an increase in the similarity of mission statements between the two sectors, which seems to be driven by for-profit organizations’ mission statements becoming more similar to non-profit organizations’ mission statements. We further explore whether the increase in similarity can be explained by the market orientation displayed in the mission statements, the moral values present in the mission statements, and the SDGs that the mission statements relate to.