Introducing textual analysis tools for policy informatics: a case study of e-petitions

Electronic petitioning (e-petitioning) provides a unique and promising channel through which people can directly express their policy preferences. E-petitions may be viewed as a natural laboratory for determining subjects of public interest, and thus can be used by policy analysts to understand social needs and constraints. In this paper, we introduce textual analysis tools (such as NER and topic modeling) and extract three types of novel variables (informativeness, named entities, and 21 topics) from We the People petition texts. The regression result shows that informativeness, named location, and several topics are significantly correlated with the log of the signature counts. These exploratory but promising results indicate that textual analysis tools can complement traditional statistical methods by providing descriptive measures that are helpful for making causal inferences from electronic petition data. These new tools, we believe, will facilitate policy analysis and policy informatics by enabling meaningful use of large volumes of online archives containing public expression regarding policy preferences.