The use of field experiments in political science has become extensive, but the promise of conducting a randomized intervention in a “real world” setting also raises perils for researchers. Partnering with organizations to deliver a randomized intervention may be a cost effective route to data collection, but a long-distance partnership presents certain challenges. In particular, the researcher needs to be especially vigilant about treatment application given the potential for noncompliance with the random assignment schedule. I provide an evaluation of a field intervention’s effectiveness where the long-distance partner organization inadvertently violated the random assignment of voter precincts in a canvassing effort prior to a citizen vote to repeal a non-discrimination housing ordinance protecting LGBT residents of a Midwestern city. I then provide recommendations for researchers to help mitigate treatment noncompliance when they cannot be present during treatment delivery.