Click for next page ( 64


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 63
The Uses of Social arid Behavioral Research Just as our everyday lives have been transformed by computer chips, communications satellites, and advances in medical technology, so, too, have they been transformed by public opinion polls, standardized tests, demographic projections, and human factors engineering. Moreover, our notions about ourselves and each other-about racial differences, for example, or the nature of childhood have been radically transformed by the dissemination of social and behavioral research findings. This chapter reviews a few of these developments. We stress once again, however, that a comprehensive account covering the entire range of such developments is beyond the scope of this or any single report. The chapter is divided into three sections. The first reviews three inventions that are used to generate information necessary for planning, analysis, and decision making: sample surveys, standardized tests, and economic models. The second section reviews a variety of changes in practice resulting from behavioral and social research findings. The third section reviews the way social and behavioral research contributes to conceptual change, altering the ideas and understand- ings we have about ourselves and our society. INFORMATION-GENERATING TECHNOLOGIES A major contribution of the social and behavioral sciences has been the invention and development of information-generating technologies. Much has been written about the information revolution, made possible by the development of high-speed computers. Equally important, perhaps, is the 63

OCR for page 63
64 BEHAVIORAL AND SOCIAL SCIENCE RESEARCH: PART I ethos of our society that places high value on informed decision making and informed public discussion of alternative possibilities. Given that we are a large and diverse society with complex economic, political, and social institutions, extensive information is needed to inform decisions and to analyze their effects. It is difficult to imagine how either government or business could function these days without economic indicator data. De- pendence on a variety of social and political data is nearly as great. Accordingly, there has been and continues to be great impetus for the development of improved information bases and improved decision-making procedures, and basic research in the social sciences has played a central role in these developments. Here we review three technologies developed by social and behavioral scientists: the sample survey, the standardized test, and econometric forecasting models. SAMPLE SURVEYS Perhaps the single most important information-generating invention of the social sciences is the sample survey, developed over the course of the past 50 years by statisticians, sociologists, economists, political scientists, psy- chometricians, and survey specialists. Sample surveys consist of the collection of data in a standardized format, usually from a probability sample of a population. Major advances in the technology of data collection and in probability sampling procedures have made it possible to collect systematic data on the behavior and attitudes of the American people as a whole, and on specific subgroups, with great efficiency and at relatively low cost.2 These advances include the development of household sampling frames and cluster sampling techniques; the elaboration of the standard personal interview to questionnaires, telephone interviewing, and structured personal diaries; the HA probability sample is one for which there is a specified probability that each case in the population will be included in the sample. The most basic kind of probability sample is a simple random sample, in which each member of the population has an equal probability of being chosen for the sample. Usually, more complex designs are used. For example, because there are fewer blacks than other racial groups in the population, blacks are sometimes sampled at a higher rate than others to ensure that the sample includes the minimum number of blacks necessary to achieve reliable estimates. In addition, sometimes samples are designed with multiple stages to reduce data collection costs: First cities are sampled, with their probability of being included in the sample proportional to their population size; then sections of the city are sampled; then blocks within sections; then households within blocks; and finally persons within households. 2Suppose, for example, that we wanted to estimate the proportion of the population of the United States that is Roman Catholic from an item on religious identification in a national sample survey of 1,500 people. If we were to take many simple random samples, each of 1,500 people, and if the true proportion of Roman Catholics were 25 percent, then 95 percent of all

OCR for page 63
The Uses of Social and Behavioral Research 65 sampling of administrative records, past and present; and data collection in experimental settings, especially large-scale social experiments (see the paper in Part II by Tanur for a review of some of these developments; see also Deming, 1968, and Stuart, 1968, on sampling>. As a result, the sample survey has become for some social scientists what the telescope is to astronomers, the accelerator to physicists, and the microscope to biologists- the principal instrument of data collection for basic research purposes. This is not to say that sample survey procedures are without difficulties. There is currently lively debate in the research literature about the reliabilty and validity of survey responses and about the efficacy of alternative procedures; see, for example, the National Research Council report Survey Measurement of Subjective Phenomena (Turner and Martin, 1981~. The value of the sample survey extends beyond its use by social scientists. It has become the major source of information for a wide variety of public and private purposes, and as a consequence has become a highly marketable technology. Most information collected by federal, state, and local govern- ments for planning and policy-making purposes depends on sample survey methodology. The statistical systems of most industrialized nations, which provide information on health, housing, education, welfare, commerce, industry, etc., are constructed largely on the methodology of sample surveys (see Keyfitz, 1968; Hauser, 19751. One form of data of particular value for information purposes is statistical time series, in which measurements of a given phenomenon are made repeatedly at successive points in time monthly, quarterly, annually, de- cennially, etc. We now have fairly long series for a number of social and economic phenomena, some running 100 years or more but most dating from the 1950s or later. The accumulation of time series data enables us to discern important social and economic trends and hence to monitor social change. In addition to a variety of well-known economic indicators (e.g., the gross national product and the consumer price index, among many others), good time series data are available on the test performance of high school and college students, health and disease rates, labor force participation rates, trends in labor force structure, and subjective phenomena, such as levels of of the samples would have an estimated proportion of Roman Catholics between 22.8 and 27.2 percent. Clearly the size of the error in any given sample is unknown if the true proportion were known, we wouldn't need a sample survey. However, the variability over all possible samples can be estimated from the variation in a single sample survey of a given size. Larger samples yield more accurate results. For example, for samples of 6,000 people, 95 percent of all of the samples would have an estimated proportion of Roman Catholics between 23.9 and 26.1 percent. Thus, increasing the sample by a factor of 4 decreases the interval of results by a factor of 2.

OCR for page 63
66 BEHAVIORAL AND SOCIAL SCIENCE RESEARCH: PART I alienation, trust in government, and racial attitudes, to name only a few. Such data not only provide information for policy deliberations but also alert us as to what needs attention if we are to understand social change. Sample surveys are also widely used in the private sector, where market research has become an indispensable tool to discover consumer preferences for everything from cars to candidates. No major election is now conducted without extensive polling of the electorate, a specialized form of market research. Public opinion polling has also emerged as a crucial part of the apparatus of modern journalism a way for newspapers and news magazines to take and report the pulse of the nation. If the pervasiveness of public opinion polling is, like television, a mixed blessing, it is also, like television, an integral part of modern life. One particularly important analytic use of sample survey data is making population projections. Based on techniques developed by demographers (see the paper by Menken and Trussell in Part II), projections are now routinely made of the future size and characteristics of the population of neighborhoods, cities, regions, and the nation as a whole. Commercial enterprises make heavy use of such data in decisions about where to locate offices, factories, and retail outlets and about the kinds of products to produce and the nature of the marketing strategy to follow. Local governments make similar use of such data in decisions about where to locate schools, hospitals, and other facilities and in planning for the future. The ability to anticipate increases or decreases in the tax base, school enrollments, the demand for bilingual education, homes for the elderly, the balance of social security contributions and payments, and so on all depend on population projections based in part on sample survey data. Applied survey research directly depends on methods and techniques developed in the course of basic research in the social sciences. Indeed, it would be fair to say that survey research techniques have advanced parallel to the advances of empirical social science over the past four decades. To be sure, there was extensive interaction between basic and applied concerns from the earliest days of serious survey research in the 1930s, when both Roper and Gallup began their polls and when Paul Lazarsfeld, the father of survey research as we know it, initiated his research program at Columbia University. Almost all of the methodological and conceptual advances came from university research centers-the Bureau of Applied Social Research directed by Lazarsfeld at Columbia, the National Opinion Research Center at the University of Chicago, and the Survey Research Center at the University of Michigan. The history of voting research (reviewed in the paper by Converse et al. in Part II), in particular, is almost inseparable from the history of survey research. Similarly, microeconomics is almost entirely dependent on survey data (see the paper by Heckman and Michael in Part

OCR for page 63
The Uses of Social and Behavioral Research 67 II), as is the study of social stratification and mobility in sociology and the study of attitude formation and attitude change in social psychology, to name just a few. One of the things social scientists use sample surveys for is to establish empirical regularities in attitudes, behavior, and social processes. Such empirical regularities-for example, the stable pattern of age-related mortality in all societies, the differential voter turnout between presidential and off- year elections, the relative invariance in markedly different societies of hierarchies of occupational prestige, the division of labor by gender, and the lower earnings of women in all industrial societies provide the building stones for theories of social structure and social process and constitute information that can inform choices made by individuals and organizations. For example, age-related mortality rates have important implications for public health programs in developing societies, and differential voter turnout has implications for campaign strategies and electoral reform. STANDARDIZED TESTING Next to the sample survey, probably no other invention of the social sciences has had as far-reaching an impact on American life as the standardized ability test. While systematic testing procedures have had a very long history (the civil service examinations in imperial China are a classic example), stan- dardized quantitative tests based on psychometric scaling principles are a relatively recent development. Toward the end of the 19th century, psychologists began to show serious interest in individual differences and their measurement. While much of the early work concentrated on the measurement of simple sensory and motor functions, there was considerable interest in the measurement of mental functions such as memory, judgment, and intelligence. Most early efforts to devise mental tests proved unsatisfactory, however, for want of a metric for calibrating "intelligence" and other mental functions. A major breakthrough was achieved in 1908 by the French psychologist Alfred Binet, who developed a test to assist the French Ministry of Public Instruction in identifying retarded schoolchildren (Binet and Simon, 19081. The key to Binet's test is that it measured individual development against a standard of "normal" develop- ment for children of the same age, thus solving the calibration problem. The measure of intelligence used, the intelligence quotient (or IQj, was defined as IQ mental age 100 chronological age

OCR for page 63
68 BEHAVIORAL AND SOCIAL SCIENCE RESEARCH: PART I Thus, for example, a child of 10 who could answer questions and solve problems as well as the average child of 13 would be said to have a mental age of 13 and hence an IQ of 130. The Binet test was adapted for use in the United States by Lewis Terman of Stanford University and became known as the Stanford-Binet, the most famous of all tests of intelligence (Terman, 1916). - Perhaps because the early 20th century was a period of great social change in the United States, in which the need for "objective" assessment procedures was widely felt, ability and achievement tests of all kinds were rapidly and widely adopted. Aptitude tests were devised to assist in the classification of students in order to channel them into various educational tracks. Achievement tests were devised for the certification of student performance and the evaluation of instructional programs. Tests of both general abilities and specific skills were developed for use in employment screening, and similar tests were created for the classification and allocation of army recruits. Standardized testing was given additional impetus during World War II by the need to classify draftees efficiently, and still later by the need of colleges and universities to select from among an excess of applicants created by the postwar baby boom. Today the use of such tests has become so widespread that there are probably few Americans who have not had substantial experience with them. From its earliest beginnings the testing industry has relied heavily on psychometric theory and methods, and as psychometric research has pro- gressed standardized tests have become more sophisticated. Of course, as in so many other areas, the process was interactive much of the impetus for the development of psychometric techniques stemmed from the need to improve testing procedures. Developments have occurred in two major areas: the statistical theory of measurement and practical procedures for test construction. As an example of the first area, procedures have been devised for assessing and improving the reliability and validity of tests. Reliability indicates the extent to which a phenomenon can be measured without error; usually this is assessed by ascertaining the similarity of results yielded by alternate forms of a test. Validity indicates the extent to which the test measures what one thinks it measures. There are different types of validity and reliability, but one example will suffice. The purpose of college admissions tests ordinarily is to distinguish between those who are likely to be successful in college and those who are not. The test can be said to be valid insofar as those who score well do better in college than those who score poorly. It can be said to be reliable if different subsets of questions yield similar rankings of candidates, the implication being that scores are not strongly affected by the idiosyncrasies of particular questions.

OCR for page 63
The Uses of Social and Behavioral Research 69 Practical test construction procedures also have been improved substantially. For example, group testing has largely replaced individual testing, thus improving efficiency. Much attention has been devoted to the development of procedures for calibrating or "norming" tests (age has been largely abandoned as a standard, since many abilities have been shown not to vary systematically with age). And there is greater understanding of how the test- taking situation can affect performance. Contemporary standardized tests have three main uses: selection, classifi- cation, and diagnosis. Although these uses shade into each other, they are analytically distinct. A major use of standardized tests is to select from a group of candidates those best suited to an activity and most likely to succeed at it. Admissions tests and personnel selection tests are examples. A second use is for classification. Selection shades into classification when the purpose is to match people to one of several activities or programs. The use of IQ tests to place students in "fast," "regular," and "slow" classes is an example, as is the use of aptitude tests by the army to assign recruits to various training programs. The use of achievement tests to identify a student's areas of strength and weakness and to monitor remedial programs is an example of the third use, diagnosis, as is the use of vocational interest inventories to counsel individuals regarding the suitability of alternative career choices. It is probably fair to say that the impact of standardized testing has been generally positive, contributing to more informed decisions about educational and vocational choices, to more efficient allocation of people to activities for which they are well suited, and hence to the enhanced realization of human potential. Arguments supporting this view are to be found in Cronbach (1970), Buros (1972), and Anastasi (1976), which provide comprehensive summaries of the literature assessing the contribution of standardized testing. Standardized testing is not without its critics, however, and its history is not without blemishes. Although the intent of Binet's work in France had been to identify the mentally retarded in order to provide remedial instruction and by so doing to improve mental performance, Binet's conception of human intellect as malleable did not survive transatlantic passage. The testing movement succumbed to the hereditarian and racist assumptions endemic to early 20th-century American culture. Ethnic group differences in mental abilities, documented by the wholesale testing of army recruits during World War I (Yerkes, 1921), were taken as evidence of innate national differences in ability (rather than as a straightforward reflection of differences in average length of residence in the United States and average levels of schooling) and were used as the intellectual basis for the National Origins Act of 1924, which established immigration quotas. Beginning in the 1930s an environmentalist ethos began to replace the

OCR for page 63
70 BEHAVIORAL AND SOCIAL SCIENCE RESEARCH: PART I hereditarian one, and group differences in test performance came to be generally recognized as reflecting features of the intellectual environment. At the individual level, however, testing was seen as a tool for promoting meritocracy, a device for identifying talented members of the society and for efficiently allocating society's intellectual resources to various educational opportunities and occupational slots. Still later the question was raised as to whether group differences in average test scores reflect not only environmental differences but also bias in the tests themselves, a tendency for tests to underestimate the capabilities of members of disadvantaged minorities. The tension between the positive functions of tests and their potential bias with respect to particular groups continues today; it is reviewed in a recent National Research Council report, Ability Testing: Uses, Consequences, and Contro- versies (Wigdor and Garner, 19821. An important finding of that report is that ability tests are not biased against minorities; that is, on the whole they predict the performance of minorities as well and as accurately as they do for members of the majority population. The report recognizes, however, that the absence of test bias does not exhaust the questions of fairness raised by test use. ECONOMIC DATA AND ECONOMIC MODELS Despite obvious political differences about how to cure economic ills, most economic decision makers both inside and outside the government share a reliance on a set of economic indicators and decision tools that are the product of economic research conducted since World War I. It is hard to imagine how the economies of the United States and other industrialized nations could function today without the kind of economic information that is now routinely available. Economic Data Contemporary economic data are characterized by their integration into a coherent framework representing the functioning of the economy as a whole. The basic idea that the functioning of the economy depends on the interaction of its parts was introduced by John Maynard Keynes in The General Theory of Employment, Interest and Money (19361. Prior to the work of Keynes, economic series were created largely independently of one another and hence were difficult to interpret in relation to one another. Stimulated by Keynes's theoretical work, economists in the United States and Europe developed a comprehensive set of economic indicators covering all aspects of production, consumption, saving, and investment for the economy as a whole. Important work was done by Kuznets (1941) and others on national income accounts, \

OCR for page 63
The Uses of Social and Behavioral Research 71 showing transactions among these sectors. This is the work that gave rise to the concept of gross national product (GNP) and many other familiar economic indicators. In a related development, Leontief (1941) invented a way of analyzing the inputs and outputs of specific industrial sectors of the economy in relation to one another. Meanwhile, work had been progressing on the theory of index numbers, with a major contribution by Fisher in 1922. An index number measures the magnitude of a variable relative to some specified level of the variable. Thus, if we take the price of a candy bar in a given year as our standard, and it costs 25 cents that year and 30 cents a year later, the index for the cost of the candy bar in the second year would be 120 (= 100 x 30/251. Index numbers for single commodities present no problems, but considerable complication is introduced when one aggregates several commodities whose prices are changing in different ways, as for example in the construction of the consumer price index. Fisher worked out a consistent set of principles for carrying out such aggregations. The result of these efforts has been the generation of a large number of economic time series that are mutually consistent and analytically coherent. An adequate data base was a necessary precondition to the development of models of the functioning of the economy as a whole; such models have been widely used both for forecasting short-run changes in the economy and for assessing the impact of various proposed economic policies. Econometric Forecasting Economies like that of the United States are highly complex and relatively volatile, fluctuating as a result of the interaction of specific events and policy manipulations. Given this, it used to be extremely difficult to predict short- run changes in the economy with accuracy. While it still is not possible to predict such changes with complete accuracy, the development of computer data bases and of large-scale models of the U.S. economy (and now of other economies as well) has substantially improved prediction, so much so that economic forecasting has become a substantial industry both in the United States and abroad, with a commercial market exceeding $100 million in the United States (Klein, 19801. Econometric macroeconomic models consist of dozens to hundreds of equations that must be solved simultaneously (one model consists of more than 2,000 equations), each equation representing the interrelationship of particular facets of the economy. For example, one equation in the model might specify how the level of future consumption depends on changes in the level of disposable income of consumers, while another equation might specify how changes in disposable income depend, on the average, on levels of national output, taxation, etc. These models are used to predict the future

OCR for page 63
72 BEHAVIORAL AND SOCIAL SCIENCE RESEARCH: PART I course of the economy by projecting trends in various economic indicators and by using the model to estimate the effects on them of alternative policy measures and of still other variables. The models are also used to predict the response of the economy to such perturbations and shocks as increased oil prices, military action, taxation, manipulation of the money supply, worldwide speculation in raw materials, etc. The sorts of responses that are predicted are changes in inflation, interest, and unemployment rates, among others. The utility of these models was established early when they began to yield predictions that ran counter to, but proved more correct than, the conventional economic wisdom of the time. A crude model constructed by Klein in 1946, for example (Maugh, 1980:758), contradicted widespread predictions that the postwar economy would fall into another depression. The standard view at the time was that there would be as many as 6 million unemployed people in this country. Klein's model, however, indicated that there was a large pent-up demand for consumer goods among the civilian population and a large amount of cash available to returning soldiers, suggesting that the economy would thrive. That proved to be the case. A slightly more developed model later correctly predicted that there would be only a modest recession following the Korean War, while many other economists were again predicting depression. Over a dozen major firms now provide econometric modeling services, and some have been providing quarterly estimates for over 20 years. These services are purchased by industrial corporations, financial firms, government offices (federal, state, local, and international), and foreign companies. Moreover, many models have been developed for foreign countries, and efforts are under way (through Project LINK, sponsored by the Social Science Research Council) to create a model of the world trade system by integrating macroeconomic models of the economies of individual countries and regions, including some eastern European countries. Efforts are just beginning to be made to incorporate the People's Republic of China into a world model. These developments have depended heavily on basic research on macro- economic processes, and almost all of the developmental work necessary to create the models the theory of system design, the methods of estimation, the methods of testing, the techniques of simulation, and data management procedures-has been done in economics departments and research centers of universities. Related Developments Modern production methods taught in business schools are the direct outgrowth of optimization theory developed through basic research in economics. Linear programming, scheduling theory, inventory control theory, quality control

OCR for page 63
The Uses of Social and Behavioral Research 73 models, production forecasting models, etc., used extensively by businesses, the military, and almost all other large organizations, are all the direct descendants of yesterday's basic research in economics. But production is not the only area of business that has been profoundly affected by basic research. Modern finance and accounting both reflect the theory of decision making under uncertainty that was a major academic innovation a few years back. And portfolio management techniques have revolutionized the opera- tions of banks and related financial institutions. CHANGES IN THE WAY WE DO THINGS Concrete applications of findings from basic research in the behavioral and social sciences can be divided into two main categories: those concerned with human performance or the design of human environments and those concerned with the structuring of social, economic, and political systems. The former category mainly includes applications of psychological theory and research, principally psychophysics and learning theory. The latter category encompasses organizational analysis, applications of location theory, economic and demographic analysis, and the evaluation of economic and social programs, among others. Many of the latter applications involve the identification and analysis of problems rather than the creation of solutions. This is because most issues having to do with social and economic arrangements do not admit of technical solutions; they require instead political solutions, that is, adjudication among competing interests and conflicting values. The role of research in this context is to provide the framework within which debate can proceed, by identifying the issues involved and providing the information necessary for their resolution. In this section we review a few examples from both of these categories. Again, our examples are illustrative; we have made no attempt to be comprehensive. HUMAN FACTORS APPLICATIONS Human factors engineering is a branch of technology that is concerned with the application of theoretical and experimental psychology to human-machine systems. There are vast numbers of such systems, the designs of which can be judiciously directed by the results of cognitive research. For example, how should a control panel be designed so that the operator can most easily remember the operation of numerous dials and switches? The poor design of the instrument panel at Three Mile Island has been said to have been an important contributor to the accident there. How should aircraft instrument

OCR for page 63
82 BEHAVIORAL AND SOCIAL SCIENCE RESEARCH: PART I and climatic amenities. These developments were incorporated into new models, which in their turn proved to be capable of providing reasonable predictions and general guidelines for siting decisions based on comparative cost methods. The most profound challenge to locational analysis emerged with greater public recognition of the effects of siting decisions. Everyone cannot be equally accessible to desirable facilities, such as workplaces, bus stops, libraries, fire stations, and hospitals. Furthermore, everyone cannot be equally remote from nuclear power plants, toxic waste deposits, solid waste disposal plants, drug treatment centers, and group homes for delinquents. To take account of preferences for accessibility to or distance from desired or noxious facilities, respectively, it was necessary to elaborate the models used to determine optimal sites. In particular, normative concepts were needed for balancing the virtues of efficiency with those of equity and justice. Analysis thus turned to the identification and measurement of neighborhood spillovers, the negative externalities that are most prominent close to a site and diminish with distance from it. To cite an example, a new hospital in a residential area displaces apartment dwellers to provide a residence for nurses. The apartment dwellers are relocated but must pay a larger share of their income for rent. The additional traffic, parking, and noise from ambulances detract from the quality of life for those who are left. The hospital serves and benefits a large portion of the city, yet its costs are borne largely by local residents. Similarly, nuclear power plants impose severe costs to land use in their immediate vicinity. Moreover, a serious accident would affect the population living within a large radius around the facility. New urban highways impose noise and some hazard to neighboring areas. Neighbors of sites selected for group homes for the mentally handicapped, delinquents, or drug abusers voice complaints about added hazards, undesired neighborhood change, and declining property values. There has been some success in measuring these unintended neighborhood spillovers, including both tangible factors and many of the intangible ones. Comprehensive impact analyses can be carried out to evaluate alternative sites, with one option being to build no facility at all. These analyses provide the bases for assessing the true costs of projects and facilities. Basic research applicable to the solution of these complex and controversial siting problems is not limited to location theory and welfare economics. Basic contributions in political theory have been found to be of great relevance in understanding the mechanisms of threat and conflict that are often the outcome of imposed siting solutions. The same body of literature is helpful in developing negotiation and arbitration mechanisms for involving citizen groups in the resolution of siting conflicts.

OCR for page 63
The Uses of Social and Behavioral Research PROGRAM AND POLICY EVALUATION 83 Another important area of application of basic research in the social and behavioral sciences is the analysis of actual practices, programs, policies, and plans. Sometimes the analysis is retrospective, to decide what went wrong, as in the case of the Three Mile Island nuclear reactor accident. Organizational sociologists and human factors psychologists who were members of the study commission were able to identify flaws in both organizational and human factors design. This sort of analysis is quite analogous to what engineers do when a bridge falls or an airplane crashes. Often analysis is undertaken of ongoing programs or activities, to understand them better and perhaps improve them. Periodic educational assessments and management reviews are cases in point. Finally, there are analyses of proposed future courses of action or policies, to anticipate their consequences. Procedures for analyzing specific programs have been formalized under the rubric of evaluation research. But even social and behavioral scientists who do not identify themselves as evaluation researchers often engage in this kind of activity. As we have noted, the analysis of alternative economic policies is a major part of the work of many economists. More and more, the work of social scientists is employed by congressional research agencies such as the Congressional Budget Office and the Office of Technology Assessment. And many basic researchers from time to time find themselves addressing quite specific problems with practical implications. Evaluation techniques emerged directly from the development of research methods in a number of the social and behavioral sciences, in particular psychology, survey research, and economics, in which serious attention has been paid to the development of sensitive assessment methods to determine experimental outcomes in complex settings. As a result, a sophisticated set of analytic procedures has been transferred to nonacademic settings. One particularly interesting evaluation study is that reported by Frederick Mosteller in his 1981 presidential address to the American Association for the Advancement of Science (Mosteller, 1981~. He studied 28 social, medical, and technological innovations, each designed to provide some specific benefit or improve some specific outcome; in another study (Gilbert, McPeek, and Mosteller, 1977), 36 medical innovations based on randomized clinical trials were studied. In both studies, Mosteller found that only about half the innovations yielded positive benefit (most of the others had little effect and a few were actually harmful). Such findings are of crucial importance as a corrective to the all-too-frequent tendency to implement proposed innovations on the strength of their plausibility alone, a tendency especially pronounced among policy makers with respect to social innovations.

OCR for page 63
84 BEHAVIORAL AND SOCIAL SCIENCE RESEARCH: PART I In some instances the federal government has asked social scientists for research explicitly designed to aid in the advance evaluation of public policy alternatives. A prominent example of such research is the large-scale social policy experiment. As the federal government considers major reforms in health, housing, and welfare policies, it has sponsored experimental programs to evaluate in advance the effect of potential national programs. Among these experiments have been programs for income maintenance (Heclo and Rein, 1980), health insurance (Orr, 1980), and housing subsidies (Field, 19801. Similar research projects, on a somewhat smaller scale, have assessed programs for employment training, criminal rehabilitation, and preschool education. Concepts and methods from the social sciences have been used in designing these studies, collecting the data, and interpreting the results. A different kind of evaluation research involves the examination of an ongoing program, to decide whether to continue it or how to improve it. A case in point is the Head Start program. In the early 1 960s several comprehensive reviews of psychological research suggested that early. life experiences are crucial in the formation of intelligence and in comprehensive development (see especially Hunt, 1961; Bloom, 1964~. This research led to the inference that children whose early life experiences were impoverished would be permanently handicapped in their cognitive development, and hence in their educational achievement, and also to the hypothesis that early intervention in the form of intellectually stimulating and supportive pre- school experience-could give such children a head start in life that would permanently improve their intellectual performance and life chances. These ideas, developing at a time when the war on poverty was being fought through a variety of social and economic programs, led to the creation of Project Head Start, a national program of compensatory preschool education, and to a large number of similar locally based programs. Many of these programs were designed as experiments, to test whether appropriate preschool education could increase IQ and improve school performance. The history of this research is illuminating, cautioning against premature generalization about the effects of social intervention. The first results were extremely encouraging. Several studies found a sharp increase in the IQ of children enrolled in the programs when compared with children not enrolled. However, a second round of studies came to much gloomier conclusions, finding that IQ and performance changes were often short-lived, disappearing a year or two after the children had left the program. While one interpretation attributes the latter finding to the continuing influence of school environments, noting that both the experimental and control groups tended subsequently to experience the same, largely inadequate elementary schooling, the more general interpretation was that compensatory education has no lasting effects. This conclusion in its turn proved premature, however, as evidence began

OCR for page 63
The Uses of Social and Behavioral Research 85 to accumulate regarding the long-term effects of early interventions. While permanent IQ changes did not in general occur, children exposed to the preschool programs were less likely to require special education, less likely to be held back in grade, and more likely to complete high school; they also tended to score somewhat better on academic achievement tests and to have higher educational aspirations (Brown and Grotberg, 1980; Schweinhart and Weikart, 1980:62-64,75-871. While data continue to be collected and the efficacy of preschool interventions continues to be debated, these interventions would not have been conceived without the large body of basic research on the effects of early childhood experience, and they could not have been evaluated without the procedures derived from the methods of experimental social psychology. Often policy makers have neither the time, resources, nor inclination to design a study relevant to a specific policy choice. In such circumstances social scientists are frequently asked to review the relevant research literature and provide informed judgments about the implications of existing findings for the policy under consideration. This task is difficult and delicate because the criteria for generalizing from such findings are stringent. The task illustrates the importance and closeness of the relationship between policy research and the development of theory and methods within the social science disciplines. Policy scientists must combine conceptual and methodological skills from a diversity of disciplines and must organize these resources, usually within a short time, to address policy issues. Because the quality of policy research ultimately depends on the quality of the basic research skills and knowledge available, the usefulness of the policy sciences depends on the development of the social sciences more generally. CHANGES IN THE WAY WE THINK ABOUT THINGS As individuals and as sharers of a culture, people develop shared understand- ings about the world around them, including the social world and human nature. Not only do these understandings reflect basic cultural assumptions, but they may also be shaped and transformed by systematic knowledge. In the United States empirically based and verified knowledge recently has come to play an important role in shaping our society's understanding of itself and of human nature. The consequence has been a fundamental transformation in the way we think about ourselves and our social arrangements. Most obviously. human behavior and social structure are now widely perceived as orderly, explicable, and amenable to systematic empirical investigation. Moreover, social and behavioral research has dramatically altered the understanding of the informed public regarding a host of social phenomena, and it promises to continue to do so. For purposes of illustration, consider

OCR for page 63
86 BEHAVIORAL AND SOCIAL SCIENCE RESEARCH: PART I how our conceptions of race and ethnicity and their correlates and our understanding of unemployment have changed as a result of social and behavioral research. THE CHANGING CONCEPTION OF RACE AND ETHNICITY The transformation of conventional wisdom on the nature of race and ethnicity over the course of this century can readily be traced through a comparison of successive editions of the Encyclopaedia Britannica. The 11th edition, published in 1911 and purported to be a comprehensive summary of available knowledge (Koning, 1981), was quite clear in its conclusions about "the Negro race" (in the 11th edition the world's population is divided into three distinct races; article on "Negro," p. 3441: In certain . . . characteristics . . . the negro would appear to stand on a lower evolutionary plane than the white man, and to be more closely related to the highest anthropoids. Mentally, the negro is inferior to the white. The remark of F. Manetta, made after a long study of the negro in America, may be taken as generally true of the whole race: "The negro children were sharp, intelligent, and full of vivacity, but on approaching the adult period a gradual change set in. The intellect seemed to become clouded, animation giving place to a sort of lethargy, briskness yielding to indolence. We must necessarily suppose that the development of the negro and white proceeds on different lines. While with the latter the volume of the brain grows with the expansion of the brainpan, in the former the growth of the brain is on the contrary arrested by the premature closing of the cranial sutures and lateral pressure of the frontal bone." This explanation is reasonable and even probable as a contributing cause; but evidence is lacking on the subject and the attest or even deterioration in mental development is no doubt very largely due to the fact that after puberty sexual matters take the first place in the negro's life and thoughts. By the 14th edition, published in 1929, the inherent mental inferiority of Negroes had come into question, although the comment about their standing on a lower evolutionary plane was retained. Carr-Sanders is quoted as concluding that "there seems to be no marked difference in innate intellectual power. The differences are rather differences in disposition and temperament, . . ." and the judgment is offered that "given suitable training, the Negro is capable of becoming a craftsman of considerable skill, particularly in metal work, carpentry and carving" (article on "Negro," p. 1931. The article on "Differential Psychology" observes (p. 368) that "the inferiority of the negro to the white in mental capacity has often been asserted as the result of comparative studies, but it is difficult to say how much of the difference found is due to native against cultural factors"; it then goes on to note that "the greater the admixture of white blood, the closer does the negro approach

OCR for page 63
The Uses of Social and Behavioral Research 87 the white in performance...." The 1929 edition is agnostic on the question of racial differences in character, noting that "whether in the final analysis there is . . . a greater tendency toward crime inherent in the Negro race is a question upon which there is great difference of opinion and little scientific data" (article on "Negro, the American," p. 196~. By the publication of the 15th edition in 1974, a dramatic transformation had taken place. There is no article on "Negroes," and the article on "Minorities and Ethnic Groups" is an analysis of the social structuring of ethnic relations and the behavior of ethnic groups in multiethnic societies. On the question of racial differences in intelligence, differences in test scores are acknowledged as typical but are discounted as almost certainly a reflection of environmental differences. It is worthwhile quoting the relevant paragraphs of the article on "Intelligence, Distribution of" (pp. 676-677~: In studies of racial differences in intelligence in countries where white culture is dominant, it has been a very consistent finding that groups classified as Negro or black are likely to achieve lower scores on standard intelligence tests than do groups of whites or Caucasians. Yet, despite typically significant differences between the mean scores of the two so-called racial groups, the ranges (i.e., the spread between the lowest score and the highest score in each group) are usually found to be about the same, revealing extensive overlap in the score distributions of the two groups. Indeed, even in studies in which average scores strongly seemed to favour whites, considerable numbers of black subjects achieved higher scores than those of the average white. Such studies have been sharply criticized since in most of them all subjects were tested by white examiners. In studies made to assess the effect of the examiner's race on the subject's scores, black children tended to earn higher scores when they were tested by black examiners than when they were tested by whites. Examiner effects should be controlled in seeking to discover possible racial differences in intelligence test scores; otherwise the results are likely to be misleading. Black children in Northern cities of the U.S. tend to exhibit higher IQs than do black children in Southern cities; the scores achieved by Northern black children who have migrated from the South tend to increase with the length of time they have lived in the North. Controlling for such influences serves to diminish apparent racial differences considerably. Yet, when tests designed to be relatively free of cultural and language biases are used, smaller racial-group differences appear. Such differences still may reflect environmental effects upon the development of the skills necessary to achieve high scores on intelligence tests, since representation of blacks in the lower social strata throughout U.S. society in the 1970s was disproportionately large. Thus, while it has been argued that IQ data constitute support for those theories of genetic racial inferiority, it seems more likely that the differences reflect persistent social and economic discrimination. Under legal or de facto segregation in many countries, educational facilities for blacks rarely have been equal to those provided for whites (e.g., South Africa, Angola). There is considerable evidence that the growth of intelligence, as reflected in the IQ, is influenced significantly by the quality

OCR for page 63
88 BEHAVIORAL AND SOCIAL SCIENCE RESEARCH: PART I of available educational resources regardless of race. For example, lower class whites also show lower average IQ scores than do middle class whites in European and North American countries. How can we account for these changes in perspective' While many forces were involved, it is fair to say that research on racial differences and institutions was a significant contributor. First, many studies have been done showing that racial differences in achievements can largely be explained by differences in opportunities and environments. Perhaps the classic study in this genre is Klineberg's (1935) demonstration that while Northern whites had higher average aptitude scores than Northern blacks and Southern whites had higher average scores than Southern blacks, Northern blacks scored higher on the average than Southern whites. There have been many subsequent studies of a similar type, some mentioned in the passage from the 1974 edition of the Encyclopaedia Britannica quoted above. Much of this material was cited by the Supreme Court in its 1954 decision declaring separate schooling for blacks unconstitutional. Work of this kind continues today. Second, our racial institutions have been subject to careful scrutiny by social scientists. The classic study is that by the Swedish sociologist Gunnar Myrdal, who was invited by the Carnegie Corporation in 1937 to head a comprehensive study of the race problem in America. Perhaps because Myrdal was a distinguished foreign social scientist deliberately chosen to provide an outsider's perspective, and perhaps also because he was able to enlist the aid and support of the bulk of American social scientists doing research on race relations and on the black population, his 1944 report, An American Dilemma, was extremely influential. This massive volume (1,483 pages) provided a comprehensive review of the social, economic, and political status of Negroes in the United States. But its real impact was due to the position it took regarding the cause of the inferior status of blacks- that it was not the peculiarities of blacks themselves that explained their position. but the history and continuing presence of discrimination against them by the white majority. The "dilemma" of the title was the discrepancy between American ideals of fairness, justice, and equality of opportunity and the reality of the treatment of the Negro minority. This report stands as a hallmark of the kind of contribution social research can make to societal self-awareness. Third, we can now monitor changes in both racial differences and racial attitudes, thanks to the development of statistical time series. We know, for example, both that the objective position of blacks relative to whites has improved substantially in the past 40 years, with a reduction but not complete elimination-of the gap in educational attainment, occupational achievement, and income (Farley, 1977), and that white prejudice against blacks has been markedly reduced (for example, the proportion of whites

OCR for page 63
The Uses of Social and Behavioral Research 89 who thought blacks are as intelligent as whites rose from 40 percent in the mid-1940s to 80 percent by the early 1960s; Hyman and Sheatsley, 1964:69. An argument can be made that the very revelation of attitudes such as a belief in the inferiority of blacks contributes to their reduction, through the process of publicly labeling them as prejudices, especially when combined with a growing scientific literature attributing race differences to features of the social environment. Finally, it has come to be recognized, through the work of both cultural anthropologists and social psychologists, that the concepts of race and ethnic group are themselves socially structured. Racial and ethnic distinctions have little or nothing to do with genetic differences but rather are social categories invoked as the basis of group identity and group differentiation. The kinds of racial or ethnic distinctions invoked and their salience vary according to historical circumstance and social setting. The Black Power movement and the subsequent resurgence of white ethnic identification, for example, are explicable as manifestations of intergroup competition for limited resources. Equally dramatic alterations of cultural conceptions resulting from research in the behavioral and social sciences could be documented for child development? aging, sex differences in behavior, mental illness, alcoholism, and the nature of foreigners and foreign cultures. It is fair to say that as a result of such research Americans today have a very different view of human behavior and social institutions than their parents did a generation ago. THE FUNCTION OF SOCIAL SCIENCE LABELING The labeling process has been one of social science's most useful functions. Social scientists need to label human activities, processes, or events so that they can bound and isolate their subject for study. The public, too, needs to be able to fix human affairs by labeling them in order to examine an issue from many sides and to consider it in detachment from its ongoing, often controversial, context. By identifying issues and highlighting their most generalizable, salient features by giving them names, social scientists assist the public in thinking through its position on alternative solutions to social problems. A case in point is juvenile delinquency. The label itself identifies the phenomenon as something different from adult crime and suggests that it may have different causes and different solutions. Similarly, inflation- the reduction in the value of money is conventionally thought of as a concomitant of rapid expansion of the economy. The term stagflation was invented to describe a different and new phenomenon the reduction in the value of money in a stagnating economy. Again, the term itself suggests the need to look for causes and cures other than those conventionally associated with inflation.

OCR for page 63
90 BEHAVIORAL AND SOCIAL SCIENCE RESEARCH: PART I Sometimes efforts to reconceptualize by redesignating go awry, as new concepts repeatedly regain their unwanted, usually negative, meanings. Consider the search for appropriate ways to categorize societies that have not attained the levels of commercial and industrial development of the modern West. Over the past century social thinkers and social scientists have used a succession of terms for these societies savage, primitive, backward, nonliterate, preliterate, undeveloped, underdeveloped, less developed, de- veloping, and finally, in a kind of desperation, simply new-only to find each term picking up connotations of inferiority. We have also had trouble referring to the very poor in society, seeking euphemisms like the hard-to- reach and problem families, only to find these labels failing as well. From these examples it is evident that labels are useful and tend to gain currency when they illuminate rather than obscure the phenomenon or the category they refer to (Matza, 1966~. Today our language is full of social science labels, terms whose meanings have moved from the common language to social science, which then have returned to ordinary use bearing the special meanings that social science has fixed them with. A short list of such words might include: depression, inflation, human capital, the hidden economy, unconscious, reference group, status, standard of living, quality of life, sample, acculturation, socialization, alienation, and unemployment. To illustrate the dynamics of labeling, the history of one label that is, one concept is an instructive example: the word unemployment. Unemployment is a venerable word, first recorded in the Oxford English Dictionary from John Milton's use of it in Paradise Lost, "Other Creatures all day long Rove idle unimploid, and less need rest." Throughout the colonial period and the early national years of the United States, the word meant something not being put to use, idle. For a spindle or a plow to be idle thus meant at most waste, but when idleness was applied to people it carried strong moral overtones. For the Puritans idle hands were the devil's playground ("For Satan finds some mischief still for idle hands to do," Isaac Watts, Against Idleness and Mischief, and for l9th-century Americans idleness was a potent sign of moral weakness. As a consequence a great deal of blaming of the victim attended the frequent depressions of the 19th and early 20th centuries when thousands of working men and women were thrown off their jobs and were unable to find alternative employment. In a nation of farmers there was no lack of work to be done on the farm. In time, however, we became a nation of cities, factories, and large businesses and, consequently, of financial panics and market collapses, bread lines, throngs of men at the factory gates, and idlers and loungers on the porches and sidewalks of the cities. Despite such scenes, the popular notion that unemployment was self-chosen idleness and moral laxity foreclosed successful

OCR for page 63
The Uses of Social and Behavioral Research 91 public debate over possible remedies for widespread suffering. What was needed in the 19th and early 20th centuries was some way to separate the personal condition of idleness from the social condition of market-induced unemployment. Social science contributed to that separation. Economists, social reformers, sociologists, and others began in the late 19th century to use the term unemployment in what we would now call descriptions of labor markets and in connection with new fields, such as the study of business cycles and monetary fluctuations. As the 20th century progressed, these technical meanings of unemployment-lack of work caused by fluctuations in the demand for labor attached to the old word, and it became possible to debate employment policies. Concurrently, as research advanced in the field, ancillary terms, likefull employment, were coined. Indeed, so defined, full employment became a possible goal for public policy: In 1946 the United States Congress passed its first Full Employment Act. Now that the social science meanings have become firmly affixed to the term unemployment, it is possible to look back to see the consequences of this redefinition of the term. First, a large number of Americans have been liberated from undeserved stereotyping and abuse, as people have increasingly come to distinguish between large-scale social and economic events that create unemployment and the individual choice of idleness. Second, social science knowledge about employment and unemployment has led to new questions, new subjects, and the coining of new words. The simple questions of when is a person unemployed and how unemployment is to be measured have turned out to be immensely complex social and economic subjects. For example, before the Great Depression of the 1930s, government statistics on the work force were maintained on the number of people with a gainful occupation. But with this approach it was not possible to make a satisfactory count of the number of people who wished to work but could not find jobs. To fashion policy during the 1930s, the government needed a count of the number of people seeking jobs. The concept of gainfully employed was replaced by a new concept, the labor force, which included everyone in the population age 14 and older (later 16 and older) who, during a specified period (usually the week prior to the date of data collection), actually had a job or was seeking work for pay. Thus the working-age population could be divided into those with a job, those unemployed and seeking work, and those not in the labor force (retired, keeping house, going to school, institutionalized, etch. This concept has guided the collection of labor market statistics ever since. Not only has it permitted the monitoring of the unemployment rate (which is now done on a monthly basis by the U.S. Bureau of Labor Statistics using data from the Current Population

OCR for page 63
92 BEHAVIORAL AND SOCIAL SCIENCE RESEARCH: PART I Survey, a sample survey conducted by the U.S. Bureau of the Census) but it has also provided better estimates of labor force participation rates for various demographic groups. Of great interest is the long-term trend toward a convergence of the levels of labor force participation of men and women: The rate for women is increasing and is now over 50 percent, while that for men is declining and is now about 75 percent. Research in the 1980s suggests there is a need to adjust economic and demographic statistical series again. Social scientists are now documenting the existence and apparent permanence of an informal economy and system of production. Some estimates, for example, suggest that unreported income could amount to 10 percent of the gross national product in the United States and even more in other countries, such as Italy. Research on the informal labor market helps to clarify patterns of intranational and international labor migration. It casts issues of unemployment in a new and sharper light. It also raises questions about the accuracy of national measures of economic growth and productivity. Recent evidence indicates an explosive grown in employment in the service and light manufacturing sectors, much of it hidden from the present accounting system (for a review see Ginzberg and Vojta, 19811. It is research of this sort that can lead to a reconception and improvement of the national reporting system. Finally, the multiplicity of social science meanings that attach to the word unemployment as research proceeds does not foreclose, but rather informs, public debate about employment and unemployment. Should mothers of infants work outside the home for wages? If they work part time, or if they quit for two years and then seek to return to paid work, are they unemployed? Should teenagers be working? Are they, in the language of the 17th century, idle "greet boys" who are liable to become rogues and vagabonds? Should families be coaxed or coerced into moving from regions of high unemployment to regions of plentiful jobs? Such questions all gain accuracy and concreteness from social science research, but none can be answered except by reference to public values and the political process.