@techreport{ai26-1521, title = {{Conditional Hypothesis Generation for LLM-Based Text Analysis with Researcher-Specified Covariates}}, author = {Paiheng Xu and Jing Liu and Wei Ai}, institution = {Annenberg Institute at Brown University}, type = {EdWorkingPaper}, number = {26-1521}, year = {2026}, month = {July}, doi = {10.26300/knbg-hb51}, url = {https://edworkingpapers.com/ai26-1521}, abstract = {A core goal of computational social science is to discover interpretable differences in how language varies across outcomes of interest, such as political affiliation or instructional quality. Recent LLM-based hypothesis generation methods describe such differences in natural language, but select for globally discriminative patterns without accounting for covariates — identified through researchers' domain knowledge — that shape the data. When covariates are ignored, selected patterns can reflect confounds rather than differences of substantive interest. We introduce conditional hypothesis generation, a framework that incorporates researcher-specified covariates to steer hypothesis discovery toward differences that hold within relevant subgroups. Two challenges arise: the target subgroup may be underrepresented (stratum imbalance), and the direction of a difference may reverse across subgroups (sign reversal). We propose two econometrics-inspired methods: one introduces feature–covariate interactions to detect sign reversals, and the other applies within-stratum demeaning and inverse-frequency reweighting to equalize underrepresented strata. Synthetic experiments show each method outperforms global baselines in its targeted setting, and expert evaluation on two real-world datasets confirms that covariate-aware generation surfaces more useful hypotheses within relevant subgroups.}, }