The main question addressed in this paper relates to the quality of the data collected via mobile telephone. We ask if it is possible to overcome the difficulties that wireless telephones present in order to collect data of substantive value. Three data sets will be compared. The first two consist of results from identical RDD surveys conducted nationally in 2003, the first carried out by cellular telephone and the second by conventional telephone. The third set comes from the replication in 2003 by the Pew Research Center of their landmark 1997 response rate experiment. To judge by response rates-a standard measure of quality, we would not have much confidence in the matching landline and cellular studies, especially the cellular survey. Both suffer from response rates in the twenty to thirty percent range. The rigorous Pew survey achieved a comparatively high rate of fifty-one percent. Thus it will serve as the gold standard by which we assess the other two surveys. The analysis will focus on the relationships among variables. Preliminary research suggests that few marginal differences in attitudinal and behavioral measures across the three surveys reach statistical significance. However, no study has extensively examined differences in relationships either by mode or by response rate. The matched landline and cellular surveys replicated attitudinal questions from the Pew response rate experiment. For each of these items we first describe the variables that theory and research indicate have the greatest explanatory power. We then apply an appropriate multivariate model to the Pew data. In the last step, we fit the same multivariate models to data from both the cellular and landline surveys. Our purpose is to explain how well these results conform to the theoretical models, paying particular attention to what any differences may reveal about the quality of data collected via mobile phone.