The benefits and problems of online surveys are well documented and this electronic data collection tool continues to increase in application among leading market research companies and public opinion pollsters in the U.S. and worldwide.  Concerns about low response rates and self-selected participation biases, sample representativeness and data generalizability have led to the application of various methodological techniques to test and adjust for sample representativeness as well as analytic strategies to assess outcome and online surveys continue to grow in methodological rigor.
The present study focused on a four-year longitudinal study beginning with a census of a targeted population of young adults followed by a random sample of this cohort the following two years and another census in the fourth year.  Its objective was to track the impact of a health promotion campaign that was implemented in the first year of the study and which continued over the next three years.  Questions in both surveys included health knowledge, attitudes and behavior as well as health information sources. 
Fieldwork was conducted during four consecutive falls among students at a northeast midsized university.  They comprised a survey group-administered through gateway courses comprising a census of entering first-year students in 2005 and a random sample of this student cohort in their second and third years followed by another group-administered survey of senior courses in the fourth year.  University students are considered an Internet-wise population and all of them have a university email address.  Research has shown that they may be effectively incentivized to participate in surveys.
Methodological issues are discussed.  These include extrapolation to the larger population based on results from specific demographic segments, whether different incentives and survey invitation wording contributed to respondent bias, the implications of leverage salience theory, latent confounding effects of students majoring in disciplines with higher sample representation, and the rigor and limitations of group-administered and online surveys.