In lieu of an abstract, here is a brief excerpt of the content:

CHAPTER 12 Caveat Populi Quaestor: The 1992 Preelections Polls in the Israeli Press GABRIEL WEIMANN University of Haifa Three heated and close campaigns were conducted during 1992: British, American, and Israeli elections were all held within a period of six months (April to November 1992). All three campaigns attracted worldwide attention and massive media coverage. All of them were accompanied by public opinion polls and surveys, monitoring the standing ofparties and candidates. The prevalence of the polls was inspired by the extremely close contests between Labor and the Conservatives in Britain, the Likud and Labor in Israel, and between Bill Clinton and George Bush in the presidential race in U.S. However, despite the popularity of the polls and their prominence in the mass media, the results were rather disappointing: the British polls failed to predict the winners, experiencing the most disastrous result in the history of polling in the United Kingdom.' The Israeli polls, though accurately predicting a change of government, failed to predict the gap between the leading parties and the resulting coalition. Two decades ago Philip Meyer suggested that '~ournalists would be wrong less often if they adapted to their own use some of the research tools of the social scientists" (Meyer 1973, 3). Sixteen years later, in spite of growing criticism of his notion of the emergence of "precision journalism," Meyer kept his optimism: "the media have made considerable progress in understanding and interpreting election polls ... The proliferation of polls will, of course, force journalists to become more proficient at interpreting them" (Meyer 1989, 1). However, the emergence of "precision journalism" 255 > Home | TOC | Index 256 Weimann was not as fast growing as the popularity of mass-mediated polls. This tendency was noted by Bogart, who claimed that "public misunderstanding of opinion surveys can be expected to continue as long a~ the mass media ignore or belittle their technical intricacies" (Bogart 1972, 23), and by Wheeler, who argued that "the reader ofopinion polls is given no hint that they may not be trustworthy" (Wheeler 1976, 16). A study of poll reports in the U.S. press led researchers to conclude: 'The way methodological information about polIing is reported in the media tends more to reassure than alert the audience about tlle possible defects of poll data" (Paletz et al. 1980, 506). The present chapter exanlines the way preelection polls were reported by the Israeli press during six election campaigns. The analysis focuses on the methodological and statistical problems revealed in press reports on polls. While most studies of this kind measure the accuracy of the reported polls by comparing their predictions with the actual results, the present study highlights the errors made in presenting, analyzing, and interpreting polls results tllroughout the campaign. I. Data Data for this study come from a systematic content analysis of fifteen daily newspapers published in IsraeP during the election campaigns of 1969, 1973, 1977, 1981, 1984, 1988, and 1992. The coders were trained to analyze any report on polls published during the three months preceding each election, and they used a preset manual of coding categories (see Charlebois 1979; Stone and Morrison 1976; Weimann 1983a, 1983b, 1990). The following components were recorded for each report (N= 1531): 1. Form of coverage: space allocation, placement in the paper, and source of sponsorship. Space was measured by column inches devoted to reporting a poll and its results and to analysis or interpretation of polls. 2. Content of coverage: inclusion of information on sampling metllOd and sample size, wording of questions, rate of "don't know" and "undecided" answers, method of interviewing, timing, measures of intensity of opinion, identification of pollster, and tlle prediction made by the poll. 3. Style of coverage: "horse-race" polls (focusing on predictions, personal popularity) or issue-related polls (focusing on issues and attitudes) . The coders were trained prior to the investigation and were subjected to an inter-coder reliability test. This test involved the comparison of the evaluations of different coders when analyzing the same text (Holsti 1969, 135-42). > Home | TOC | Index Caveat Populi Quaestor 257 The coefficients of reliability (i.e., the average proportion of agreement between pairs of coders) were, for the 1992 data, .92 in form, .89 in content, and .92 in style. Data for previous elections came from our database on press reports of preelection polls, updated every campaign (Weimann 1990). ll. Reporting the Polls The Proliferation ofPolls When looking for the origins of election polls, many turn to 1936 when Gallup, Roper...

Share