AccessMyLibrary provides FREE access to millions of articles from top publications available through your library.
EMERGENCE AND FIRST WAVE OF PROLIFERATION
The survey method is the single most important approach in empirical social research - at least in the United States. Some time ago, a study of leading social science journals (Presser, 1985) showed that survey data were used in 55.8% of all articles in the three leading sociology journals during the 1979-1980 period - up from 24.1% in 1949-1950 and on level with 54.8% in 1964-1965; in political science, the share of articles based on survey data increased from just 2.6% in 1949-1950 to 19.4% in 1964-1965 and to 35.0% in 1979-1980.(1) I am not aware of any recent replication of this analysis, but a cursory review of the journals analyzed by Presser does not suggest any dramatic changes. Survey data do provide the main base for systematic empirical analysis of many important topics in sociology, political science, and related fields. To avoid any misunderstandings, given the wider use of the term survey in everyday English and the unfortunate tendency to guise a variety of operations as surveys,(2) a clarification on the use of this term in this article may be helpful. A survey is defined as the collection of data for the purpose of scholarly inquiry by use of a standardized questionnaire administered by specially trained interviewers (in person or over the phone) or distributed (predominantly by mail) to a (randomly) selected sample of respondents for self-completion. Furthermore, I will focus on surveys used to collect data on mass sentiment (attitudes, beliefs, opinions) and behavior with respect to social and political issues - sometimes referred to as attitude surveys.(3) Political should be understood in a broad sense, going beyond just voting intentions, party preferences, or the popularity of political actors.
The modern survey method is intimately linked to the development of random-sampling techniques for large human populations, which started in the 1930s and 1940s, notwithstanding the fact that some roots of the survey method can be traced back to at least the last century (see, e.g., Converse, 1987; Marsh, 1982). Two key studies - very different in nature - set the course for the survey method to become the method of choice in the United States. The first was the poll conducted by George Gallup at the occasion of the 1936 presidential election, the second the comprehensive set of studies in social psychology in World War II, commonly known as the American Soldier study (Stouffer et al., 1950). Gallup's main contribution was the merging of random-sampling techniques with data collection via interview, demonstrating that even a relatively small sample of maybe 1,000 respondents can adequately represent the attitudes, opinions, and behavior of a population as large as 100 million. The simultaneous failure of the Literary Digest poll in 1936 has become a standard example, used in almost every single textbook on methods of social research. The other study (American Soldier) was crucial in sharpening and developing the tools of questioning beyond the relative restricted realm of political polling. After its completion, many scholars involved in this project moved to the University of Michigan and founded the Survey Research Center (SRC) (Rossi, Wright, & Anderson, 1983). As part of a larger Institute for Social Research (ISR), the Michigan center became and still is the Mecca for survey researchers worldwide.
From the United States, the survey method spread to other Western industrialized countries such as Great Britain, Germany, Japan, or France, though with various speed, and often met with a deep-rooted suspicion of being positivist - with positivist often used as a shorthand for simplistic, affirmative, unimaginative, manipulative - and lacking the sophistication of social analysis in various philosophical traditions (see, e.g., Marsh, 1982). More sophisticated arguments included a linkage of functionalist theory with the survey method, but even this assertion is controversial and has been challenged by, for example, a fact-based comparative case study of Harvard and Columbia (Platt, 1986). In a nutshell, American social science clashed with European social philosophy. For decades, empirically oriented sociologists were deeply divided into a quantitative camp derisively labeled bean counters and a qualitative camp belittled as able to produce at best journalistic accounts with no scientific value. The ebbs and flows of the academic debate on the survey method varied from country to country. In (West) Germany, for example, this unproductive debate reached its climax in the late 1960s and the early 1970s, and then dissipated in the 1980s. Today, the survey method has become a major, although not always the dominant, part of academic social science in Western Europe.
Although the academic world was reluctant to embrace the survey method, entrepreneurial spirits in both private industry and (government-related) organizations were quick to recognize its potential and became pioneers in establishing the survey method in their countries. Two examples illustrate this more general pattern. In Germany, Elisabeth Noelle-Neumann founded the "Institut fur Demoskopie" shortly after the end of World War II. Her institute quickly became one of the leading survey research institutes in the country focusing on monitoring public opinion on a wide range of issues and often using innovative techniques in questionnaire construction. Noelle-Neumann, of course, later gained an international - though somewhat controversial - reputation as a scholar, but her impact on the academic realm in her home country has always been rather limited. Another example of the proliferation of the survey method via nonacademic institutions is Japan. In 1946, the NHK Broadcasting Culture Research Institute was established to "conduct surveys and studies on general broadcasting from the perspectives of culture." Four years later, its mission was expanded "to periodically conduct scientific public opinion surveys and to publish the results to ascertain the wishes of the public" (NHK, 1996). Projects include the Election Public Opinion Survey, the Survey on Current Affairs, the Survey on Value Orientations of Japanese, and the Survey on Lifestyle of Japanese - going way beyond the narrow realm of media research. Similarly, the office of the prime minister in Japan has sponsored an annual survey on social attitudes since 1950 (Sasaki & Suzuki, 1990).
PUBLIC OPINION AND DEMOCRACY
Commercial research firms (including worldwide Gallup affiliates) were established in numerous countries, which heavily used the survey method and provided the general public with an image of itself, with information on mass sentiment toward a variety of issues. In particular, political polls, surveys about politics and elections, brought high visibility - acclaim as well as criticism - to these commercial institutes. The survey method and the (quantitative) study of public opinion have been intimately linked from the survey method's very beginning in the 1930s to this very day. And the relationship between political polling and democracy is a perennial topic of sometimes heated debate (for an overview, see Crespi, 1989). At face value, the survey method is genuinely democratic; everyone has the same chance to be heard (to be selected into the sample), everyone's opinion counts exactly the same, nobody exerts pressure on the respondents to answer in specific ways (the interviewers are trained to act as friendly, but neutral and accepting, recorders of whatever opinion or answer is offered), and in the end, the results are widely distributed, and everyone is free to draw his or her own conclusions. No dominance by elites, no group pressure, no spin doctors. The survey method simply provides a mirror to the public and helps to foster a comprehensive dialog in an open society.
Sounds too good to be true? Well, of course. We all - and specialists in survey research methodology probably more than anyone else know that survey data, the responses given, are the product of a rather complex information-processing process that is affected by a multitude of internal and external factors. By external, I mean deliberate attempts to skew the results to serve a political, social, or religious agenda as well as problems arising from lacking and/or inferior skills to carry out a survey project; by internal, I mean factors beyond the control of an unbiased scholar on top of the state of the art in survey methodology and all its aspects including sampling, questionnaire design, questionnaire administration, and data analysis. No honest survey researcher will deny that survey data have their shortcomings, more apparent in some studies than in others, no matter how hard one tries to capture the "true" attitudes, opinions, feelings, and behavior of the people in the sampled population. This is not the place to review the multitude of texts on how to do surveys, but one of the most interesting and intriguing lines of research into survey methodology is the focus on the cognitive processes involved (e.g., Schwarz & Sudman, 1996; Sudman, Bradburn, & Schwarz, 1996; Tanur, 1992), although its impact on day-to-day survey research, especially when done under difficult conditions, is still limited.
So, although the survey method undergoes constant refinement in countries with a (relatively) long tradition of survey research, its international appeal is significant and hardly diminished by methodological problems with this approach, even in its "native" countries. This is documented in both the geographical expansion of a growing number of cross-national survey projects and the proliferation of the survey method to countries beyond the Western democracies of North America and Europe.
CROSS-NATIONAL SURVEYS: STANDARDS
Cross-national survey projects provide a seemingly easy way to export the survey method to countries with relatively little experience and familiarity with this method. After all, one important part - the conceptualization and operationalization of questionnaire contents - has been done and has been proven successful in countries with longer traditions of survey research; for other parts, such as sampling and survey administration, norms and procedural rules have been established. Adding to the popularity of expanding cross-national projects is the fact that the potential for gaining significant insights is greatly increased by the availability of comparative data. Rarely does a one-shot survey provide a secure base for further-ranging conclusions; replication over time and in different contexts vastly enhances the value of any survey project. It may be next to impossible to eliminate all systematic bias from survey data, but in many cases, it is safe to assume that the factors producing such systematic bias are relatively stable over time (in a given context). Consequently, when making comparisons over time, there is a good chance that any systematic bias cancels out. Furthermore, by comparing over time, looking at trends rather than at measurements at one point in time, fluctuations due to (random) sampling can be distinguished from substantive change.
Cross-national survey projects replicated in regular intervals, then, have proven extremely popular, as the success of the International Social Survey …