Because of their cost/time-efficiencies, web surveys are being increasingly incorporated into national survey data collection programs in the United States. Yet, coverage and data quality issues in web surveys remain important challenges. As a basic study designed to better understand data quality in a mixed mode national survey, this paper investigates the (1) socio-demographic characteristics of web vs. mail respondents, and (2) the effects of survey mode on response rates and data quality in a 2008 Gallup Health panel survey. The Gallup panel examined is selected with probability-based sampling from the U.S. general population and employed to conduct a mixed mode web-mail survey (number of respondents = 39,444). Whether panelists complete a web or mail questionnaire is pre-assigned according to their internet-use patterns, which were identified at the time of recruitment.

Findings indicate that the mode selection procedure produces significant socio-demographic differences between web and mail respondents. Web respondents’ characteristics were found to be significantly different from those of mail respondents in terms of age, gender, race, household-income, education level, urbanity and the duration of panel participation. After multivariate adjustments for these differences in sample composition, mode effects on response rates continue to be documented. Although the web survey mode produced significantly lower response rates overall, compared to the mail survey mode, respondents in the youngest age group were nonetheless more likely to respond when assigned to the web survey mode. In addition, when examining data quality via open-ended response patterns, some evidence was found indicating that the web survey mode may produce higher quality information, although multivariate controls suggest these mode effects may be limited. Our presentation will include consideration of future research needed to improve the quality of mixed mode surveys that include a webbased component.