Web-based surveys are attractive to researchers for a variety of reasons. The extent to which these results can infer population level characteristics is largely limited by sample design and non-response bias. This project seeks to explore the impact that email contact language has on college students likelihood to complete a survey. Because college students have free access to internet, are usually required to have a university email address, and often receive correspondence through these email addresses they are an attractive population for internet surveys. Unfortunately, because of this it is more difficult for any one request to stand out enough to get attention. This study expands on findings by Rath et al presented at the AAPOR 2015 conference that contact emails with humor have significantly higer response rates than more traditional formal language. The study includes a random selection of community-college students from all 13Tennessee Board of Regents Community Colleges (~60,000 students will be included in the study). The larger purpose of the study is to examine non-medical prescription stimulant use among this population. Students will be randomized into one of two invitation conditions – formal and informal language. The initial contact language will be mimicked through all subsequent contacts. Our hypothesis is that students will be drawn to the informal language because it more closely mirrors how they write and speak but we are also interested in determining if the informal language may increase response among prescription drug abusers. Data collection will be completed by the end of March 2014 and the analysis will look at response rates by treatment, patterns in non-response, patterns in rates of use by treatment condition, and demographic differences. This project seeks to add to the body of literature surrounding survey recruitment methods for web-based surveys, particularly for those that involve sensitive questions.