My research is oriented around the development and evaluation of methods for computational social science. My basic contention is that the field of NLP is well-served served by grounding it in the needs of social science—giving direction to concepts like interpretability and generalization. On the methods side, I'm interested in identifying latent constructs (perhaps call it "very abstractive summarization"): topic models, ideal points models, NLG for annotation. In terms of evaluation, I like to think about validity: are we really measuring what we want to measure? Topically, I'm drawn to work on bias & fairness and political science; more recently, I've started work on constructs in mental health, specifically those relating to suicidality.