OXFORD JOURNALS OXFORD UNHVEftSTTf PRESS American Association for Public Opinion Research Polls as Journalism Author(s): Irving Crespi Source: The Public Opinion Quarterly, Vol. 44, No. 4, Polls and the News Media: A Symposium (Winter, 1980), pp. 462-476 Published by: Oxford University Press on behalf of the American Association for Public Opinion Research Stable URL: http://www.jstor.org/stable/2748466 Accessed: 23/02/2010 11:46 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=aapor. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. 9 STOR American Association for Public Opinion Research and Oxford University Press are collaborating with JSTOR to digitize, preserve and extend access to The Public Opinion Quarterly. http://www.jstor.org Polls as Journalism IRVING CRESPI Mhis paper attempts to examine almost a quarter century's personal experience to lay a groundwork for the systematic analysis of how both the strengths and weaknesses of public opinion polls, as they function today, derive from their ties to journalism rather than from their own inherent qualities. Such analyses are necessary if effective ways for correcting existing weaknesses are to be developed. (See also Crespi, 1980; Roper, 1980; and Roll and Cantril, 1980: Preface). My personal experience has been primarily in relation to the use of polls by newspapers and magazines, with relatively limited contact with television. For this reason, I shall make only passing comments on intermedia differentiation. And since my own experience in polling began in 1956, I cannot report directly on the first 20 years (1936-1956) of modern public opinion polling. Historical Ties of Polling to Journalism Newspaper polling, specifically with respect to elections, can be traced back to the 1820s (Fenton, 1960:7). In the first three decades of the twentieth century, straw polls were conducted by a variety of newspapers, e.g., the Hearst Newspapers, New York Herald, Cincinnati Enquirer, Columbus Dispatch, Chicago Tribune, Chicago Journal, New York World, Cleveland Press, Omaha World-Herald, Des Moines Tribune, and the St. Louis Times. Magazines such as the Farm Journal, Pathfinder, and the ill-fated Literary Digest were also early sponsors of straw polls (Robinson, 1932). The year 1935 witnessed the emergence of both the newspaper Abstract Public opinion polling has always had close financial and institutional ties to journalism. Many of its strengths and weaknesses derive from these ties. Irving Crespi is currently a Vice President of The Roper Organization, Inc. He was with the Gallup Poll for 20 years. Public Opinion Quarterly Vol. 44:462-476 © 1980 by The Trustees of Columbia University Published by Elsevier North-Holland, Inc. 0033-362X/80/0044-462/$1.75 POLLS AS JOURNALISM 463 syndicated Gallup Poll and the Fortune (Roper) Poll, and in 1936 King Features syndicated the Crossley Poll. The correct forecast of Roosevelt's landslide victory over Landon by all three polls established public opinion polls as a standard journalistic feature in presidential election years. (Their status in off years, however, was much weaker). The Gallup Poll was subscribed to by over 100 newspapers, which received three news releases each week, while the Fortune Poll became a regular feature of that magazine for the next 20 years. (Other magazines such as Life, Look, Saturday Evening Post, and Redbook intermittently sponsored polls for feature articles during the 1940s and 1950s.) Also, during the 1940s, a number of media-supported state polls were organized, some syndicated to newspaper subscribers—Joe Belden's Texas Poll in 1940 and Mervin Field's California Poll in 1947—while others were run in-house by newspapers themselves—Des Moines Register's Iowa Poll in 1943 and Minneapolis Tribune's Minnesota Poll in 1944.1 Today, media sponsorship of public opinion polls has mushroomed. The Harris Survey has competed with Gallup as a syndicated news feature since the early 1960s, and national polls are now regularly conducted by CBS/New York Times, NBC/Associated Press, ABC/ Harris, the Washington Post, and the Los Angeles Times. Additionally, Time and Newsweek have been frequent sponsors of public opinion polls since the 1960s. Statewide and local polls are regular features not only of these media but of the New York Daily News, Chicago Sun-Times, Boston Globe, and more than 150 other newspapers and television stations. The 1970s saw the rise of "precision journalism," the application of the polling method to news coverage by individual newspapers (Meyer, 1979). Perhaps the most notable characteristic of this recent growth is the expansion of media-conducted polls, as differentiated from the media-sponsored poll. The result of this development is not merely that public opinion polls are, more than ever, featured news items, but that the institutional ties between polling and journalism are more tightly integrated than ever. The Role of the Editor The focus of interaction between polling and journalism is the editor. He has a decisive role in determining what the polling budget should be, how much space should be allocated to reporting poll results, and what topics to poll on. The way in which the news editor 1 Personal communication with Joe Belden, founder of the Texas Poll; Mervin Field, founder of the California Poll; and Glenn Roberts of the Iowa Poll. 464 IRVING CRESPI exerts his influence varies according to the specific organizational relations between the survey organization and the editor, so that in some cases his influence is muted and indirect while in others it is immediate. But in all instances it is decisive. The newspaper or magazine editor's role is unambiguously powerful when polls are conducted in-house or under direct commission. In polls that I directed under commission to Newsweek, I always dealt directly with the editor, who personally reviewed all questionnaires. In television, the CBS, NBC, and ABC polling units are part of their respective news divisions, and have no structural relation with other survey research units within the network.2 The individual news editor's influence is most difficult to specify in syndicated polls such as the Gallup and Harris Polls. As one of some 150 subscribers, no single editor can decide how a syndicated poll shall function. Indeed, with so many different clients, syndicated polls have an appearance of seeming freedom and independence of action. Editorial influence in these cases is to a large degree based on the polling firm's contractual obligation to produce a specified number of news releases each week, releases which the editor is free to print or not as he chooses, although under most contractual arrangements he is not permitted to alter them. That is, with few exceptions, the editor must decide whether to print the poll release as issued or else not print it at all. If poll releases are to be published, they must therefore be consonant with the editor's news judgment. Thus, poll releases must be produced on schedule, and must then compete with unpredictable news developments for the finite amount of space allocated to news. Syndicated polls cannot expect all their releases to be published, but if a large proportion are not, cancellation of the subscription is inevitable. It is consequently incumbent on the pollster to conform to the criteria of news editors in designing each survey. Newsworthiness A distinguishing characteristic of public opinion polling as it has evolved under media sponsorship is that the views of the general public on political issues and events are treated as the primary subject matter, worthy of attention in their own right. This contrasts with private polls sponsored by candidates and political parties, whose purpose is to identify and analyze the factors that must be taken into account in developing successful election campaigns (Nimmo, 1970; Napolitan, 1972; Perry, 1968). Rather than treating opinions as ex- 2 Personal communication with Roy Wetzel of NBC and Warren Mitofsky of CBS. POLLS AS JOURNALISM 465 planatory or intervening variables for analyzing behavior, media polls seek to investigate what the public thinks about candidates for office, prospective legislation, and other matters of public concern primarily for its news value. The results of media polls may be, and often are, of instrumental value to candidates, legislators, and other office holders. But that does not in any way change the fact that their prime objective is to report aspects of public opinion for their intrinsic interest as news. The constraint exerted by news editors on polls reflects the concept of news worthiness that dominates American journalism (Roshco, 1975; Gans, 1979; Tuchman, 1978). This constraint manifests itself in the criteria used to select poll topics and to decide what questions to ask about them. The most newsworthy expression of grass roots public opinion has always been voting, since in the normal course of events it is the only form of expression that has political clout. Compared with the lobbying activities of organized interest groups and the opinions of those in leadership positions, the opinions of the general public have limited direct effect on legislative and policy issues (Gallup and Rae, 1940; Erickson and Luttbeg, 1973; Ziegler, 1974; Boynton et al., 1969; Schriftgeisser, 1951). Therefore, the public's opinions tend to be judged by news editors as irrelevant to what is happening legislatively. In contrast, the expression of public opinion in the voting booth has an immediate, direct, and powerful political effect: therefore, it is news. Moreover, being able to anticipate correctly the outcome of an election before anyone else is a form of news scoop. This accounts in largest part for the preoccupation of the news media with preelection polls—especially as predictions—and the media's comparative neglect of polls dealing with issues. A major consequence of this preoccupation with preelection polls has been to weaken the financial stability of polling organizations in the periods between elections. For example, the typical pattern since 1936 has been for the number of newspaper subscriptions to the Gallup Poll to decline during these interim periods. These declines created a need for newsworthy topics, preferably political, to poll on during nonelection years. "Human interest" topics—such as health, vacations and leisure, and relations between the sexes—proved to be only moderately satisfactory. Most successful were political questions, notably "trial heats," which test the early voting strength of prospective candidates for the two parties, "primary tests" which present lists of potential candidates for each party's nomination; and "popularity" ratings of the incumbent president (Crespi, 1980). Focusing on political personalities rather than issues, and always pointing ahead to the next election, these questions have over the 466 IRVING CRESPI years made polling a continuing quasi-election (Mendelsohn and Crespi, 1970). In this way, definitions of what is newsworthy about public opinion have played a significant role in transforming presidential election politics from a quadrennial event into a continuing process. The linkage between journalism and polling is a key element in a feedback loop, in which (1) definitions of what is news, (2) determine the content of polling, which (3) affects the political process, (4) which then becomes news. Timeliness A key criterion used by news editors in deciding what to publish is timeliness. News is what is happening now, or is about to happen, not what happened last week—or even yesterday. Polls, it follows, should illumine today's events, be relevant to today's headlines, be on the moving edge of history—and not be merely historical documentation of yesterday. (Preelection polls excel in meeting this criterion.) If a poll is to be deemed worthy of publication by an editor, it must not appear dated. The pollster must, therefore, select topics that are unlikely to become dated rapidly. Budgetary considerations normally preclude syndicated polls from conducting a separate personal interview survey on each topic, making it necessary to make such survey an omnibus, as it were. In practice, this has meant conducting personal interview surveys every two to four weeks, with perhaps six to nine releases (e.g., two a week over a period of three weeks) to be produced out of the results of each. A serious complication is the time needed to develop questions, conduct interviews, process and tabulate the data, and distribute the results nationally. Thus, up to six weeks might elapse between the design of such a survey and the issuance of the last release—a long period over which to maintain timeliness. While telephone surveys have to some degree reduced this time pressure (see below), the problem of generating releases from one survey over a period of time persists. Because of the insistence of news editors on poll results that are not outdated, it was long the practice of the Gallup Poll not to report actual interviewing dates: Poll results based on interviewing conducted three or more weeks before publication would not be considered news by an editor, even though they were the most recent available. Up until the mid-1960s, it was standard practice to rely exclusively on such circumlocutions as "the latest" or "the most recent" poll. It was only after a good deal of soul-searching that the Gallup Poll finally decided in the late 1960s that news editors had POLLS AS JOURNALISM 467 become sufficiently knowledgeable about polls to make it safe to publish interviewing dates. Even when polls are designed to deal with a topic in depth, the journalistic need to be timely takes its toll. Responding to the violence and unrest attendant on the civil rights movement of the 1960s, Newsweek sponsored a series of pioneering in-depth national surveys of blacks—two by Harris and one by Gallup (Brink and Harris, 1964, 1967; Goldman, 1970), surveys which produced a fascinating and informative picture of black opinions. It has been over a decade since the last survey in the series was conducted, and despite continuing black/white tensions, no similar poll has since been conducted. Those few media opinion polls of blacks that have been conducted in the past decade have dealt with specific headline events—school busing and, more recently, Andy Young's forced resignation as ambassador to the United Nations.3 Avoiding the Abstract The need to be timely has a number of significant effects. For one, questions need to relate directly to specific ongoing events, rather than to broad abstractions or underlying trends. This is why media polls about politics tend to be limited to elections, opinions about political personalities in the news, or controversial proposed legislation. Basic attitudes and values underlying voting preferences and opinions on issues are typically ignored in favor of opinions regarding newsworthy events. For example, after a period of intense polling on the Panama Canal treaty (Roshco, 1978), there is a dearth of polls on U.S. relations with Panama now that the treaty has been ratified. A common journalistic technique for avoiding the abstract is to focus on political personalities in the news rather than on the underlying issues. To the extent that election campaigns focus on candidates, they are admirably suited to this technique. As one editor observed to me in a conference devoted to designing a poll on the dynamics of an election campaign, questions on issues as such produce soft, "thumb sucking" news. On the assumption that issues are subsumed—made meaningful to the electorate—by attitudes toward candidates, he wanted the poll to focus on the candidates themselves. A positive effect of designing polls around specific personalities and events is that the results relate directly to the political process. This concreteness enhances their meaningfulness in that the problems of 3 The ability of the recently established Data Black Opinion Polls to get attention from general news media will undoubtedly be dependent on how significant the black vote is expected to be in the 1980 presidential election. 468 IRVING CRESPI imputation and inference that often beset theoretical efforts to relate political behavior to underlying attitudes and values are avoided. Over time, out of the kaleidoscope of individual polls, a more structured picture of the nature of public opinion can be developed—if time and budget are available. When that happens, the theoretical study of public opinion benefits. Such secondary analyses of poll data, however, are seldom initiated by the news media themselves, and usually are the product of academicians. Journalistic considerations typically lead to a jumping from one topic to another, so that individual poll releases often present public opinion as a spasmodic, even mindless, reaction to a series of unrelated events. The Problem of Fast-Breaking News A paradoxical effect of the journalistic need for polls to be timely in the past was to create pressure for polls to avoid questions on fast-breaking news. Fast-breaking news events tend to be poor candidates for poll questions since by the time the results are ready to be released they are no longer newsworthy and are only of historical interest. Questions about events and dates that can be anticipated and planned for are much more likely to be published—for example, questions relevant to upcoming elections, the opening session of Congress, national holidays, such as July 4th, the reopening of school in September, religious holidays such as Easter and Christmas, or the beginning of a new year. Such questions can be scheduled so that release dates for reporting the poll results will coincide with the event. Of greatest interest are topics that can be expected to remain in the headlines for an extended period of time—for example, elections or hotly debated legislative battles such as over the SALT treaty. Recent technological developments and the increase in telephone households to the point where better than nine out of ten households are subscribers have enabled polls to cope more effectively with fast-breaking news events. It is now possible to design, conduct, and release a telephone poll virtually overnight. In this respect, polls conducted for television and wire services, or by newspapers that stress their national role, have a decided advantage over syndicated polls and magazine-sponsored polls. Syndicated newspaper polls have to build into their schedules time to distribute releases nationally by mail, since wire distribution is too costly for all but the most exceptional events. Magazines, on the other hand, have to cope with fixed publication dates, which often do not synchronize with news events. Wire services and television networks, in contrast, can re- POLLS AS JOURNALISM 469 lease a poll as soon as the results have been tabulated and examined. Magazine telephone polls, such as those conducted for Newsweek, on fast-breaking news events have been able to cope with this problem by scheduling interviewing for one or two days immediately preceding their press deadline. Polls on primaries illustrate the time difficulties that syndicated newspaper and magazine polls encounter. Whoever wins the first primary in an election year can expect his national poll standing to benefit immediately. Furthermore, subsequent changes in national standings can be expected when the outcome in later state primaries differ (Mendelsohn and Crespi, 1970). This responsiveness of polls to primary election results creates a problem for syndicated newspaper polls and magazine polls. A national poll conducted in the days immediately following the New Hampshire primary can be reported by a wire service or by a television network before the Massachusetts primary occurs the following week. However, a syndicated newspaper poll could not get its release to subscribers before the next week's primary without incurring extraordinary expense. Weekly news magazines, which go to press Saturday night, are able to distribute to newsstands in only selected locations before Tuesday, so that many newsstand buyers and all subscribers would not see the poll results until they have been outdated by the Massachusetts primary. Human Interest Stories One way of coping with the possibility that questions might become outdated by events is to select human interest topics that are relatively timeless. It is for this reason that questions on most admired man and woman of the year, husband-wife relations, knowledge of historical events, reading habits, places one would like to visit, dieting and interest in losing weight, ideal family size, use of drugs and alcohol, happiness, and satisfaction with one's standard of living, job, housing, and the like have long been Gallup Poll staples (Gallup, 1972). In time, some of these questions—such as ideal family size and satisfaction with life—have assumed a larger significance, and have been adopted for use by academicians, such as demographer Judith Blake (1965), and quality of life analysts. The initial role of such questions, however, was to provide material for release that was of journalistic interest and that was not likely to become outdated. In fact, results of questions like these often are saved for release until a time when, because events have reduced the newsworthiness of more timely data, there is a lack of material to report. 470 IRVING CRESPI Creating Trend Lines Another effective technique for dealing with the problem of timeliness has been to chart trends on such varied topics as presidential popularity, most important problem facing the nation, expectations regarding how prosperous or peaceful the coming year would be, amount of money a family of four needs to get along on, and candidate standings. In all these instances, the questions relate not to an event but, rather, to a specific continuing concern. Such questions, while avoiding the ever-present pitfall of becoming dated, nonetheless deal with topics at a level of immediate reality, thereby also avoiding editorial rejection on the grounds that they are too abstract to be of interest to the ordinary newspaper reader. What is noteworthy about these trend questions is that, over time, some of them—in particular, presidential popularity ratings and candidate standings—have themselves become newsworthy. That is, the release of the poll itself has become the news event! In the process of seeking ways of measuring public opinion that would satisfy journalistic criteria of newsworthiness, polls have thus become the creator of news. In the alchemy of journalism, a change in the president's popularity rating is not merely a survey result, it is something that has happened to his political strength, and therefore is news. Thus, just as they wait for the latest monthly report on what has happened to the Consumer Price Index and the Unemployment Index, editors expectantly await the latest poll report on the president's popularity. In the latter case, however, it is the news media themselves—either directly through polls they conduct, or indirectly through polls to which they subscribe—that create what they await. Objectivity Implicit in much of what has already been discussed is the idea that, according to American journalistic values, news is supposed to be objective and factual. Comment and analysis are presumably reserved for the editorial page and the columns. In this tradition, public opinion polls conducted for the media have been constrained to be impartial reporters of facts about public opinion. This journalistic constraint goes beyond conventional scientific standards that research be unbiased and objective: It requires the direct reporting of poll results without apparent intrusion of the researcher's interpretation. Exemplifying this perspective was the comment of one newspaper editor at a conference about how the Gallup Poll could best perform its reporting function. He admonished that we stick to the reporting of POLLS AS JOURNALISM 471 "hard news" like candidate standings, and leave the "soft" interpretation and commentary to his staff. In a more extreme manifestation of this attitude, Joseph Alsop (1960) castigated the Gallup Poll for not reporting its raw tabulations of candidate standings, suggesting that such analytical procedures as sample weighting and restricting the report to likely voters was in some way manipulative. This constraint to report hard, factual news has tended to create an image of objectivity regarding media polls. In contrast to advocacy polls, which are always subject to suspicion, media polls are generally assumed to be disinterested attempts to measure public opinion. Since any acts that would cast doubt on that assumption would undermine their acceptability as news gatherers, polling organizations in their work for the news media tend to be particularly sensitive to charges of bias and subjectivity (Wheeler, 1976; Kornhauser, 1946). The value that journalism places on objectivity does not, in and of itself, insure the absence of bias in either the design or analysis phase of a poll. Individual pollsters have their particular pet interests and points of view, and these are evident in the selection of particular topics and in the frames of reference within which questions are asked, as well as the specific ways in which questions are worded. However, the pollster also is aware of the fact that his poll reports will be evaluated by journalists (a breed notorious for their skepticism) looking for evidence of bias. As a form of self-protection, some polling organizations deliberately ask a number of staff members of different political persuasions to evaluate questions and analyses, hoping in this way to anticipate possible criticisms and to make suggested changes. In my experience, a sure way of getting a change in wording of questions and press releases I felt was needed was to point out how the draft form might be subject to charges of bias. Unfortunately, the constraint against subjectivity in poll reporting can lead to a devaluation of analysis. As a result, polls too often confine themselves to reporting on a limited range of topics, with no interpretation. The contrast between such a narrow view of polling and a more analytic one is evident in a November 12, 1967 New York Times report on a Huntley-Brinkley Report news special based on a nationwide poll commissioned by NBC and conducted by The Gallup Organization. NBC News was concerned that it, along with the other news media, would repeat the error of 1964 in ignoring Barry Gold-water because he was considered to be an unlikely nominee. To avoid being surprised by another Barry Goldwater, NBC wanted a poll that would focus on voter attitudes on a wide range of issues and possible candidates. As part of the poll, some standard trial-heat questions were asked. The day after the poll results were televised, the Times 472 IRVING CRESPI had a one-paragraph item on the poll, limited to the trial-heat questions and written in a style clearly patterned on Gallup Poll releases. The other, more analytical aspects of the poll apparently were not considered to be prime news. More recently, as news media have come to assign key personnel to report on polls conducted in-house, greater attention has been paid to the analysis of poll results. Space Constraints The briefness of the Times item in the 1967 instance reflects another major journalistic constraint on analysis, namely, the limited amount of space that historically has been available to polls. Only under the most unusual circumstances can a syndicated newspaper poll expect space for more than about 500 to 750 words. If the requirement that polls be factual—meaning that they report actual percentages—is to be met, there simply is no room for analysis beyond the most cursory reporting of major demographic differences. Also because of space limitations, any one release can report results of no more than two or three questions. Since each release is treated as a separate news story, the inevitable consequence is that every effort must be made to capture the essence of even the most complex issue in the fewest possible questions. This explains much of the proclivity of media polls to ask questions that are often criticized as being superficial and plebiscitarian. To meet such criticism, and to give their reports analytic substance, syndicated polls have come to rely on charting trends, and relating these trends to events (Cantril, 1944, 1967). The rigidity with which space limitations are imposed upon poll reports has recently been considerably relaxed by newspapers such as the New York Times, the Washington Post, and the Los Angeles Times—on polls that they themselves conduct. These polls are treated as feature stories, with prominent by-line reporters assigned to them. In contrast to the small weekly fee for a syndicated poll, such polls represent major financial investments for the newspapers. The result is the allocation of considerable space, sometimes a page or more over a number of days, for analyzing as well as reporting poll results. Television poll reporting has not shared in this relaxation of space limitations (even when polls are conducted cooperatively with newspapers). In the 1960s network attempts to produce news specials based on in-depth polls failed to generate audience interest. The networks then adopted the strategy of reporting their polls as straight news items. In such a format the television pollster faces a serious space obstacle when trying to make his reports analytical. It is not POLLS AS JOURNALISM 473 uncommon for a poll report to be a 30-second item on the early evening news. Two minutes may be allocated for major poll stories. Conclusion Many of the strengths and weaknesses of media-sponsored public opinion polls as they function today reflect their ties to journalism rather than qualities inherent in survey design and analysis. Surveys of public opinion can, and sometimes do, produce incisive, sophisticated analyses of the public mood. At its best, journalism affects polls by adding to or reinforcing inherent methodological qualities in four ways: 1. Journalistic requirements place a high value on factual documentation of poll results, in the form of actual percentages rather than fuzzy generalizations. 2. Subjective editorializing is devalued insofar as poll reports are concerned, reducing the likelihood that the personal views of pollsters will introduce bias. 3. Attention is focused on opinions regarding specific events and issues, thereby making poll results relevant to the real-life experiences and problems of the public, and to the political process. 4. Sensitivity to changes in public opinion, resulting from the effects of events, is enhanced. For each of these strengths derived from its ties to journalism, polling is subject to four correlative weaknesses that also derive from journalism: 1. There is a preoccupation with reporting numbers, the "objective" poll results, with a corresponding lack of interest in their underlying meaning or patterning. 2. Superficiality and lack of analysis too often characterize coverage of even the most complicated issues. 3. Topics that can be expected to create front-page headlines dominate, leading to a spasmodic coverage of the agenda of public concerns. 4. There is limited continuing coverage of long-term trends, and background news is often neglected. Dissociating public opinion polls from journalism, as some have suggested, by funding them through academic institutions or government agencies, could enable them to slough off the weaknesses that derive from journalism. However, there is a price to pay for such a dissociation, a price that might well prove to be greater than warranted. The political relevance of poll data that academic analysts have found so useful is protected and enhanced by journalistic re- 474 IRVING CRESPI quirements. If it were not for media polls on deposit at such data banks as the Roper Center, academics would not today have the extensive record of public opinion on the day-by-day occurrences of the past 45 years. Surveys conducted periodically by academic survey centers, such as NORC's General Social Survey and ISR's election studies, simply do not substitute for the frequent, regularly scheduled polls sponsored by the news media. Moreover, it is difficult to comprehend how university-based polls could fulfill this function, considering their academic obligations and the time schedules they normally follow.4 The limitations of government sponsorship in respect to this function are obvious to anyone who has experienced the nit-picking, time-consuming process known as "OMB review" (Kreps, 1978; Office of Management and Budget). This process involves the preparation of a lengthy, written statement justifying the burden the survey will place on respondents. This justification must cover such matters as the need for the survey, as well as the appropriateness and soundness of the research design. In addition to the considerable time and effort required to prepare the statement, a personal meeting with an OMB clearance officer is often required. No questionnaire can be administered to more than nine people, not even for a pretest, without OMB clearance. Four weeks is a realistic minimum to complete the process once the justification has been submitted, and it often takes three to six months. A particular hurdle for public opinion surveys is the fact that measures of attitudes and perceptions tend to be subject to special scrutiny. At least as important to protect as political relevance is the value of the journalistic tradition of objectivity. As private political polling and advocacy polling proliferate, the need increases for polls whose objectivity can be trusted. Political and bureaucratic constraints effectively exclude polls conducted by government agencies from consideration as a substitute for media polls. It is only necessary to cite two of the more obvious sources of political bias to show how great the threat to objectivity would be: congressional control over budgets, and executive concern that established policy not be endangered. Furthermore, the need to satisfy internal bureaucratic politics, in both the legislative and executive branches, and to avoid offending partisan sensibilities would preclude incisive polling on issues. Polling organizations and media pollsters must look to developments within journalism itself to shore up the serious deficiencies in 4 One exception is the Eagleton Institute's (Rutgers University) New Jersey Poll, which has been able to provide a spotty reporting of public opinion in that state, focusing primarily on candidate and party standings. POLLS AS JOURNALISM 475 polling that stem from journalism. This does not mean that professional pollsters should wait passively for such developments to occur. They do have a responsibility to stimulate them, as, for example, was done by the promulgation of AAPOR's Standards of Disclosure in 1968 and the formation of the National Council on Public Polls that same year (Mendelsohn and Crespi, 1970). In any resultant dialogue between pollsters and journalists, attention should be paid to the ways, outlined above, in which public opinion polling has benefited from its association from journalism, as well as the ways in which it has suffered. References Alsop, Joseph 1960 "The wayward press." The New Yorker, Sept. 24:170-74. Blake, Judith 1965 "Demographic science and the redirection of population policy." Journal of Chronic Diseases 18:1181-200. Boynton, G. R., Samuel C. Patterson, and Ronald R. Hedlund 1969 "The missing links in legislative politics: attentive constituents." Journal of Politics 31:700-21. Brink, William J., and Louis Harris 1964 The Negro Revolution in America. New York: Simon and Schuster. 1967 Black and White: A Study of U.S. Racial Attitudes Today. New York: Simon and Schuster. Cantril, Hadley 1944 Gauging Public Opinion. Princeton: Princeton University Press. 1967 The Human Dimension: Experiences in Policy Research. New Brunswick, N.J.: Rutgers University Press. Crespi, Irving 1980 "Polling: journalism vs. social science." In Albert H. Cantril (ed.), Polling on the Issues. Cabin John, Md.: Seven Locks Press. Erickson, Robert S., and Norman R. Luttbeg 1973 American Public Opinion: Its Origins, Content and Impact. New York: Wiley. Fenton, John M. 1960 In Your Opinion. Boston: Little, Brown. Gallup, George (ed.) 1972 The Gallup Poll: Public Opinion 1935-1971. 3 vols. New York: Random House. Gallup, George, and Saul Rae 1940 The Pulse of Democracy. New York: Simon. Gans, Herbert J. 1979 Deciding What's News. New York: Pantheon Books. Goldman, Peter 1970 Report from Black America. New York: Simon and Schuster. Kornhauser, Arthur W. 1946 "Are public opinion polls fair to organized labor?" Public Opinion Quarterly 10:484-500. 476 IRVING CRESPI Kreps, Juanita M. 1978 "Directions for the conduct of federal statistical activities." Federal Register 43, May 4:19260-65. Mendelsohn, Harold, and Irving Crespi 1970 Polls, Television and the New Politics. Scranton: Chandler. Meyer, Philip 1979 Precision Journalism. 2nd ed. Bloomington: Indiana University Press. Napolitan, Joseph 1972 The Election Game and How to Win It. Garden City, N.Y.: Double-day. Nimmo, Dan 1970 The Political Persuaders. Englewood Cliffs, N.J.: Prentice-Hall. Office of Management & Budget (no Instructions for Requesting OMB Approval Under The Federal Re-date) ports Act. Standard Form No. 83A. Washington, D.C. Perry, James M. 1968 The New Politics. New York: Clarkson N. Potter. Robinson, Claude E. 1932 Straw Votes. New York: Columbia University Press. Roll, Charles W., Jr., and Albert H. Cantril 1980 Polls: Their Use and Misuse in Politics. 2nd ed. Cabin John, Md.: Seven Locks Press. Roper, Burns W. 1980 "Impact of journalism on polling." In Albert H. Cantril (ed.), Polling on the Issues. Cabin John, Md.: Seven Locks Press. Roshco, Bernard 1975 Newsmaking. Chicago: University of Chicago Press. 1978 "The polls: polling on Panama—si; don't know; hell, no!" Public Opinion Quarterly 42:551-62. Schriftgeisser, Karl 1951 The Lobbyists. Boston: Little, Brown. Tuchman, Gaye 1978 Making News: A Study in the Construction of Reality. New York: Free Press. Wheeler, Michael 1976 Lies, Damn Lies, and Statistics: The Manipulation of Public Opinion in America. New York: Liveright. Zeigler, Harmon 1974 "The effects of lobbying: a comparative assessment." In Norman R. Luttbeg (ed.), Public Opinion and Public Policy. Homewood, 111.: Dorsey Press.