"Parishioner donation behavior is of great interest to church leaders. In many ways, parishioners behave as customers, but rather than buy, they donate. In this paper, donation habits are predicted using three well-received models of buyer behavior: (a) the beta-geometric/negative binomial distribution (BG/NBD), an alternative to the Pareto/NBD model; (b) the beta-geometric/beta-Bernoulli (BG/BB), a discrete-time model; and (c) the periodic death opportunity (PDO) model. A naïve model is also considered as a benchmark. Each model is used to predict the expected number of donations from individual parishioners. Using a novel data set of parishioner donation data, output from each model is compared and model effectiveness is evaluated using two approaches that consider a variety of giving environments. At the aggregate level, the naïve approach outperformed the more sophisticated models, but there are significant improvements at predicting donor-level behavior with the PDO model." (Abstract)