Email has enormous potential as a medium for data collection. It permits complex questionnaires to be administered more quickly, flexibly, and inexpensively than telephone or postal communications. Email, though, is currently restricted to individuals with access to computer networks, meaning we cannot generate probability samples that would enable us to statistically infer the true population parameters from our sample estimates. In recent years, scholars have begun employing email surveys to study public opinion, relying on weighting or the form‐resistant correlation hypothesis to neutralize the effects of email's coverage error. In this paper, we assess the application of these techniques, demonstrating that email surveys violate the assumptions underlying them. We discuss the implications of this for future survey research employing electronic methods.