Existing predictive models of online petition popularity largely overlooked the literature of agenda setting. This study adheres to Cobb and Elder's (1972) issue expansion model and symbolism (Birkland, 2017) in the agenda-setting literature. Examining the literature, we identified features of popular petitions and examined the effects of these features on online petition success. Commonly used models were used to evaluate our proposed features and compare their performance with benchmark

more »

... The predictive model, i.e., the product of our study, is the combination of our proposed features and the best performing model. The contributions of the study are two-fold. This study demonstrates how we can translate the textual characteristics described by the literature of agenda setting into technical features that are comprehensible to machines. On practical implications, a better predictive model helps activists better utilize online platforms to secure support for their proposed policy changes.