A potential limitation of web-only panels of the general public, even when households are selected using probability methods, is that only about 70% of households have members with Internet access. In addition, some members of Internet-connected households may be unable or unwilling to participate over the web. The Gallup Panel uses both mail and web to survey respondents. In 2006, this panel included approximately 50,000 households selected by telephone through RDD. Individuals who reported accessing the Internet more than once a week are asked to respond by the web, while all other respondents are surveyed by mail using a questionnaire with a similar visual layout to the web. The purpose of this paper is to determine whether the mail option adds value to the results. Five different approaches were used to examine whether results would remain the same or differ if mail respondents were excluded from the survey. We examine: 1) whether web and mail respondents are demographically different, 2) whether mail and web respondents provide different answers to the same questions, 3) whether traditional demographic weighting can eliminate those differences, 4) whether the addition of demographic variables eliminates the effect of survey mode in a logistic regression analysis, and 5) whether collection of data from panel members by an independent mode, the telephone, results in more accurate predictions of election outcomes from only web panelists, only mail panelists, or both. In addition, through a careful review of recent research, we explore the extent to which inherent mode differences in how people answer mail and web questionnaires could account for any differences observed, or if the current construction methods protect against that. In general, the results of these analyses provide evidence that use of mail adds value to the panel results by increasing the overall accuracy of survey results.