What happens when you need to do a web survey but your database of email records is incomplete? This paper reports the findings of a 2009 study of university alumni where a web survey was utilized. In the database from which the sample was drawn, approximately 40% of cases had invalid and/or questionable email records. In order to ensure that the methodology adopted represents all alumni in the database, a decision was made to recruit participants differently depending on the type of email record that existed for them. Recruitment, in this context, consisted of a letter and several reminders inviting those who were sampled to access a unique URL and take a web survey about their experience with the academicprogram that they had attended. Those who had an updated email record were contacted by email only (60%); those who had a questionable email record were contacted by email and by U.S. mail (28%); and those who had no email record were contacted by U.S. mail only (12%). A total of 1138 invitations were

sent out. Given the use of a mixed invitation mode, the survey field period lasted 30 days. The web survey response rates varied greatly by recruitment method: A response rate of 58% was achieved for email only invitations, a response rate of 41% was achieved for email and U.S. mail invitations, and a response rate of 12% was achieved for U.S. mail only invitations. Specific methodological details are provided in the paper and implications are drawn for future studies wishing to use a web survey method and a database of incomplete email records.