Survey research and empirical political science grew up together. Although the bills for commercial survey fieldwork are mainly paid for nonpolitical purposes, early surveys were justified publicly for their contribution to a deepened understanding of the electorate. Even today, polls on political questions are the loss leader for many high-profile firms. On the academic side, systematic quantitative investigation of political phenomena began with the Erie County Study (Lazarsfeld, et al. 1968, cited under Based on Purpose-Built Academic Data Sets), and academic and commercial practices intersected with controversies over quota versus probability sampling in the 1940s (Converse 1987). Survey research on public opinion and elections was the central force in shaping empirical methods for the discipline as a whole. Whereas survey research was initially a path along which insights from sociology and psychology were imported into political science, in time political scientists came to dominate the trade. Also with time, survey analysts were forced to acknowledge the limitations of their own method, for causal inference in general but also for historical and institutional nuance. As an expression of a scientific temperament, survey research thus yielded ground to other techniques, most notably statistical analysis of archival data on one hand and experimentation on the other. But these challenges arguably have forced the sample survey to reveal its versatility. Cross-level analyses are increasingly common—all the more so as our understanding of the statistical foundations of multilevel modeling has grown. In addition, surveys are serving increasingly as vehicles for experimentation, a way of recruiting subjects outside the laboratory and off-campus and of linking random selection of subjects to random assignment to experimental treatment or control. The current period is one of massive flux and, possibly, rapid obsolescence. On the one hand, target populations are growing less compliant with surveys, even as the bases for survey coverage become more uncertain. On the other hand, new techniques have emerged, often linked to new funding models. Most critical is the World Wide Web. Ironically, the emergence of the web as a survey platform has revived controversies, seemingly settled in the 1940s, over the requirement for probability samples. Through all of this, concern has grown about the very meaning of survey response and its relation to public opinion—indeed, if such a thing as public opinion exists.