- Shopping Bag ( 0 items )
Ships from: Mishawaka, IN
Usually ships in 1-2 business days
Ships from: North Hollywood, CA
Usually ships in 1-2 business days
Ships from: Siloam Springs, AR
Usually ships in 1-2 business days
Ships from: ypsilanti, MI
Usually ships in 1-2 business days
Ships from: Fort Mill, SC
Usually ships in 1-2 business days
Ships from: Stroud, Glos, United Kingdom
Usually ships in 1-2 business days
James S. House, Eleanor Singer, Robert L. Kahn, Howard Schuman, and F. Thomas Juster
Quantitative, empirical social science is largely a development of the twentieth century. The advent of the twenty-first century provides an apt moment for assessing its accomplishments and prospects. The present volume seeks to do this by examining the past contributions and future frontiers of one of the most significant developments of twentieth-century social science-the scientific sample survey. Our purpose is thus broad. Our examples and case studies, however, are drawn primarily from the work of the Survey Research Center (SRC) and the Institute for Social Research (ISR) of the University of Michigan, of which we are members (see the brief overview and history of SRC and ISR later in this chapter). Although we will always try to connect SRC/ISR research with research done elsewhere, readers should anticipate that much of what will be discussed here is based on work in the SRC and ISR.
We begin with the premise that social science is neither a misnomer nor an oxymoron; it is indeed a form of science, like the physical and biomedical sciences, in many of its defining characteristics. The conceptual vocabulary, the formal nature of theory, and the research methods of all the sciences vary with their subject matter. Among the natural sciences, for example, physics is arguably the oldest, having highly developed, distinct, but closely connected components of mathematical theory and largely experimental empirical research and utilizing highly sophisticated and expensive technology and apparatus. In contrast, ecological and evolutionary biology employs verbal as well as mathematical theory and heavily utilizes nonexperimental observation and measurement as well as laboratory experimentation. Chemistry and cellular/molecular biology lie in between, as do the sciences of particular natural phenomena such as astronomy, geology, meteorology, and the emerging earth/environmental sciences. All of them, however, share a respect for quantitative data; a search for general, if frequently contingent, truths; and an insistence on the continuous testing of theoretical ideas using data collected, analyzed, and interpreted as objectively as possible. All scientific disciplines develop via a complex interplay among the empirical realities and problems they seek to understand and the available theory and methods for achieving such understanding. Theory leads to new questions and hypotheses about the empirical world; empirical data lead to the revision of existing theories and the creation of new ones; and available methods and research instruments shape the limits and opportunities for empirical knowledge and theoretical development.
In the natural sciences the importance of innovations in methods and instruments is almost self-evident. For example, the telescope and its developments, from optical to radio to orbiting satellites and interplanetary probes, have continually transformed the nature of theory and data in astronomy and physics, such that we have begun to understand the origins of the universe itself. The microscope and its developments, from optical to electron to computerized imaging and genetic mapping, have similarly transformed our understanding of the nature of life, health, and disease in the life sciences, both natural (e.g., biology) and social (e.g., anthropology and psychology) (cf. Rasmussen 1997). The same can be said for particle accelerators in physics or for analytic tools in chemistry from spectroscopy to carbon dating, the latter also affecting the social science of anthropology. These innovations and their impact have also spread from science to the larger society, and many of them are now used routinely in both the private and public sectors for other than purely scientific purposes. Their broad societal impact is epitomized by the computer, first developed to address problems of basic science but now shaping all aspects of social life and much of our economy.
In the social sciences, the development of methods for scientific sample surveys constitutes perhaps the leading example of a comparable progression in methods and instrumentation, transforming both empirical knowledge and theoretical understanding. Survey methods comprise (1) precise techniques, deriving largely from the field of statistics, for selecting a relatively small sample of units from a total population (of people, organizations, or other elements) such that accurate estimates of the whole population can be derived from analysis of the sample; (2) valid and reliable measurement methods, deriving largely from psychology, for assessing attributes and behaviors of the constituent elements, usually individual people, of a population sample; and (3) methods of data processing and empirical analysis, deriving again from the field of statistics, and especially applications thereof in other social science disciplines such as economics, psychology, and sociology. Together, these methods make possible informative, descriptive, and causal understanding of social phenomena and processes from survey data.
As the title of this book suggests, we regard the sample survey as analogous to a telescope on human society (Converse 1987). The survey can provide empirical images of the populations of organizations, communities, societies, and even the world. Like a telescope, its focus can be broadened or narrowed to allow examinations of supra- and subpopulations of individuals or organizations and, ultimately, even individual communities, organizations, or persons. Prior to the development and refinement of survey methods, such comprehensive and systematic examination of social life was not possible, at least for populous and complex societies.
The chapters of this volume illustrate the impact that developments in survey research have had and continue to have on a broad range of social science disciplines and interdisciplinary areas ranging from political behavior and electoral systems to macroeconomics and individual income dynamics, mental and physical health, human development and aging, and racial/ethnic diversity and relationships. In the area of politics, survey research has drastically revised understanding of the causes and consequences of electoral attitudes and behaviors, both over the life course and across generations, and in the process has radically reshaped the discipline of political science (part 1, chaps. 3 and 4). In the area of economics, surveys have revealed the influence of psychological and behavioral factors in macroeconomic cycles of growth and recession and have also contributed to the growing understanding of the nature, extent, causes, and consequences of changes in income, welfare, and poverty (part 2, chaps. 5 and 6). The ability to systematically survey the thoughts, feelings, and behavior of members of organizations facilitated the development of theories of organization grounded in the actual rather than the assumed or imputed psychology of organizational members (part 3, chap. 7).
In the area of health, surveys first facilitated population epidemiology based on biomedical theories and then helped to transform theories of morbidity and mortality to recognize the role of social, psychological, and behavioral factors in both mental and physical health. Within SRC, ISR, and other places, the study of social aspects of health grew primarily out of studies of organizational members (part 3, chaps. 8 and 9). Systematic survey research also transformed knowledge and understanding of the causes and consequences of human fertility and family development, substance use and abuse among youth, and adolescent and early adult development more generally, and is increasingly transforming our understanding of the dynamics of aging in both individuals and society (part 4, chaps. 10-12). Since the work of Du Bois (1899) at the end of the nineteenth century, surveys have played an increasingly central role in delineating and understanding the racial/ethnic diversity of our society and the attitudes and relations among major racial/ethnic groups (part 5, chaps. 13 and 14), especially over the last sixty years. Continuing development of the methods of survey research undergird all of these developments and give promise of new approaches that will improve understanding of both human individuals and populations and of their relationship to the social and biological-physical-chemical environments in which they live and work (chaps. 1, 2, and 15).
In addition to their influence in these domains of social science, survey methods have also spawned or fundamentally transformed major enterprises in both the public and private sectors. Survey research has been the foundation of increasingly complex and sophisticated systems of government statistics within and across nations. For example, the monthly Current Population Survey (CPS) conducted by the Department of Commerce's Bureau of the Census is the source of official monthly estimates of employment and unemployment and thus of the changing state of the economy. Market research, political and public opinion polling, and organizational development and consulting have utilized survey methods to become major sectors of the economy and important influences on the behavior and decisions of leaders and groups in government and the private sector. All of this is now part of our routinely lived and assumed reality; yet almost none of it existed a century ago.
The Institutional Context of Scientific Methods and Instrumentation
Crucial to the development and maintenance of major innovations in scientific methods or instrumentation is their location within a nurturing institutional infrastructure. Smaller-scale methods and instruments (e.g., personal computers, small microscopes, and small telescopes) can be diffused across many individual offices and laboratories. However, more capital- or labor-intensive methods or instruments (e.g., mainframe or supercomputers, major telescopes, complex and expensive microscopes and other imaging devices, particle accelerators, and nuclear reactors) must be situated in larger institutional contexts, such as university- or organization-wide, regional, or even national or international facilities. Survey research can be conducted at any of these levels, but as science develops, larger-scale methods and instrumentation tend to be required for further progress. Thus with survey research, as with other major methods or instruments of science, advances in methodological, substantive, and theoretical knowledge increasingly require institutional arrangements that can sustain large-scale and long-term research programs, while allowing the broad scientific community both to influence their research designs and to utilize the resulting data. In this chapter, and again in the conclusion, we highlight ways in which supportive and innovative institutional contexts have been and continue to be necessary to the development of survey research and the scientific knowledge and innovations provided by it. Here we also acknowledge the enormous importance of broad public support for science in general and especially for survey research, which not only needs general public support but also the active cooperation and participation of the public as the respondents who are the ultimate source of survey data.
Survey research in one sense has a long history-with elements present in the registration/taxation rolls of the Roman Empire, which the Bible tells us led Mary and Joseph from Nazareth to Bethlehem, where Christ was born. Censuses, which are really surveys of total national populations, were essential elements in the development of representative democratic political systems in the United States and elsewhere. Converse (1987, 88-90) and Bulmer (2001) trace the intellectual ancestry of modern survey research to the English social surveys by Karl Marx and others in the nineteenth century (see also the foreword to this volume by Featherman); the study of attitudes and their measurement in academic social psychology in the 1920s; methods pioneered by early market research firms; and the first sample surveys of large populations in the United States by three commercial polling firms established in the mid-1930s: the Gallup Organization, the Roper Poll, and the Crossley Organization.
Several developments of importance for scientific survey methods occurred in the United States in the 1930s. Some were in the areas of measurement and the development of organizational capacity for survey research in government and academia. Rensis Likert, in his doctoral dissertation at Columbia University, demonstrated that the five-point Likert scale, as it came to be called, produced data almost identical to that generated by the theoretically elegant but cumbersome Thurstone method of paired comparisons. Likert then moved to head a research unit of the insurance industry, where he could try out both his methodological innovation and his emerging ideas about organizational leadership. In 1939 he moved to Washington, DC, to become the director of a small survey unit in the Department of Agriculture, whose mission was to conduct surveys into farmers' experiences and opinions about federal programs such as soil conservation and farmer resettlement (Converse 1987, 157). During the same decade, survey research made what is probably its earliest academic appearance in the United States. Paul F. Lazarsfeld, a Viennese émigré social psychologist, established at the University of Newark in 1935 a survey unit that, in 1940, he moved to Columbia University as the Bureau of Applied Social Research.
Even with the best and most extensive efforts at measurement, however, surveys still lacked the essential foundation of accurate and valid inferences to larger populations. This problem was epitomized by the infamous Literary Digest poll of 1936 concerning the presidential election, in which, despite a huge sample of some ten million mail questionnaires, the lists used for sampling were biased toward higher income people and the two million who mailed back their questionnaires were apparently even more biased in this direction. Literary Digest predicted victory for the Republican candidate, Alfred Landon, but Franklin Roosevelt won in the greatest landslide in American history. Gallup used a form of quota sampling that provided a better representation of the electorate and correctly predicted the Roosevelt win, providing some initial evidence of the value of the emerging science of sample surveys, though even Gallup underestimated the size of Roosevelt's victory. As noted later in this chapter and in chapter 3, the commercial pollsters proved inaccurate in the closer 1948 election, due to some combination of their quota sampling methods, the closeness of the election, and the timing of their last polls. Getting a sample of respondents to accurately represent a much larger population required the development of probability sampling, the foundations of which were also laid in the 1930s.
During the years of the Great Depression and the efforts of the Roosevelt administration to mitigate its effects, the actual number of unemployed persons became an issue of great political and economic importance. Several continuing series of unemployment estimates were published on a monthly basis, among them one by the AFL-CIO and another by the National Industrial Conference Board. Neither of these series was survey based, and their estimates were widely divergent, in directions that appeared to reflect their sponsorship. The labor unions' estimates numbered many millions; the estimates of the Conference Board were so modest that, a few years later when the economy improved, the Conference Board series began to show "negative unemployment."
Meanwhile, in the Division of Research and Statistics of the Works Progress Administration (WPA), the emerging methods of probability sampling and survey interviewing were brought to bear on the problem of estimating the national level of unemployment. In the late 1930s this new enterprise-variously titled the Nationwide Research Project, the Monthly Report on Unemployment, and the Monthly Report on the Labor Force-began to publish national estimates of employment and unemployment. In 1942 the project and its research staff moved from its home in the WPA to the Bureau of the Census, where it continues in vastly refined and expanded form as the CPS. Probably the most important early innovation in the WPA was the development of area probability sampling, largely the work of J. Stevens Stock, who later became the first sampling expert in Likert's group at the Department of Agriculture, and Lester Frankel, who moved with the WPA survey group to the Bureau of the Census. These developments built on earlier work by Keir in Norway and Neyman in England (Converse 1987, 41-45). The essential contribution was that every element of a population had a known, nonzero probability of inclusion in the sample. Statistical theory then allowed estimates (e.g., of voting intention, behavior, or employment status) for the entire population with an accurately estimable margin of error that rapidly diminished as samples increased in size from hundreds to thousands, regardless of the size of the total population (see chap. 2 for further details).
Excerpted from A Telescope on Society Copyright © 2004 by University of Michigan . Excerpted by permission.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.