An important question in legal theory and policy is when people are willing to put aside their policy preferences to uphold higher-order legal values. That is, when does constitutional or international law have “normative force”? Around two dozen experimental studies have attempted to measure this question empirically, but their designs contain an inherent limitation. Subjects in the treatment group whose priors are strongly contrary to the treatment message on the legality of a policy are effectively “treatment resistant”: it is difficult to successfully treat them because their prior beliefs on the issue are entrenched. Where a significant portion of the treatment and/or control group is not successfully treated, the results will be biased towards a null finding. This article first formally models the mechanism underlying experiments on law's normative force. Using new data from a survey experiment conducted in the United States in spring 2022 among a representative sample of U.S. residents, I then demonstrate a solution to the problem of treatment resistance. The experimental design uses a post-treatment treatment uptake test, which allows me to use the experimental treatment as an instrumental variable, which in turn allows me to measure the effect of the real explanatory variable of interest: a person's true belief about a policy's higher-order lawfulness. My insight implies that the backfire effects documented by some studies do not reflect a negative response against the legal source per se, but instead result from an experimental research design relating to hot-button issues. These findings should lead us to re-evaluate both the existing body of experimental studies on law's normative force, and how we conduct future research in this domain.