Density Dependence Without Resource Partitioning: A Population Ecology of Change.org

• In Event: Computational Methods for Studying Political Communication

Abstract

E-petitions are a prominent form of collective action. The presence
of many similar petitions suggests competition between them, yet
prior work mainly studies factors that induce individuals to
sign. We use Latent Dirichlet allocation (LDA) topic modeling to
identify topical groups reflecting ecological niches. We test
population ecology theories on \Sexpr{f(n.petitions)} Change.org
petitions. We find evidence for density dependence, an
inverse-U-shaped relationship between population size and petition
success. This suggests e-petitioning is competitive and that
initiation and participation of e-petitions draw on overlapping
resource pools. Resource partitioning theory would predict that
topically specialized petitions would obtain more signatures in
concentrated populations, but find no evidence for this. This
suggests that specialists struggle to avoid competition with generalists.
Our methodological innovation enables future such studies of textual
artifacts.

Authors

• Nathan TeBlunthuis, U of Washington

• Aaron Shaw

• Benjamin Mako Hill, University of Washington