with emotional salience and controversy. Using the random assignment of U.S. federal

judges, we employ a sparse model for estimating treatment effects with high dimensional

instruments. Democratic appointees are 10% more likely to favor permissive obscenity

standards. With this variation, we estimate that progressive obscenity standards

increase progressive sexual attitudes, non-marital sexual behavior especially by men, arrests

for prostitution, rape, and drug violations, and the incidence of invisible STDs.

To corroborate a causal channel we conduct a field experiment by assigning workers to

transcribe obscenity news reports. Exposure to progressive obscenity decisions leads to

more progressive sexual attitudes but not to self-reported sexual behavior. A second

field experiment documents that exposure to conservative obscenity decisions leads to

beliefs that extramarital sex is more prevalent. The shift in norm perception verifies a

key assumption in the model, which predicts when law has backlash or expressive effects.