Traditional, online polls reported differently

Traditional, online polls reported differently

Kim, Sung Tae

This content analysis of both traditional and online polls in selected U.S. newspapers from 1996 to 1998 finds substantial differences in the way they are reported in news stories.

In the last several decades, especially since the late 1970s, public opinion polls have been acknowledged as a necessary tool for probing the complexity of people’s beliefs and attitudes on a variety of issues. Thus, the significance of public opinion polls in news coverage has been recognized and well documented in the literature. For instance, Kathleen Frankovic, director of surveys for CBS News, said that “counting tends to make something appear more real” and numbers provide us with “a sense of definiteness.”1 Today people like to quote the latest poll when they discuss a pertinent issue. The media also love to include polls in their coverage.

Along with the proliferation of traditional polls, such as telephone and mail surveys, surveys conducted via the Internet have become more common. Both pollsters and news organizations have increasingly turned to online public opinion polls, despite criticisms of this method as non-representative and the differences between the demographics of online users and the general population. Given that an important advantage of survey research is its ability to select a sample that represents a larger population, many researchers have voiced reservations about online polling and its reporting in the media.2 Nonetheless, the number of online poll stories in the news media has increased considerably in the last several years.3

In previous studies of the use of poll data in the news media, one of the main interests has been the quality of poll reports. Given the influence of media coverage on people’s beliefs and attitudes, the question of how non-representative online public opinion polls are reported in the news media assumes increased importance and deserves to be closely examined, including a comparison with the reporting of traditional, mostly random sample, opinion polls during the same time period. To do this, this study analyzes the coverage of both traditional polls and online polls in selected U.S. newspapers during the 199698 time period to compare the frequency and the way in which these polls were reported.

About three decades ago, a number of scholars heralded the new “precision” journalism that used social science methods for news reporting.4 We now stand on the edge of another shift in which online polls are beginning to compete with traditional polls as methods for news reporting. Keeping in mind the words of Leo Bogart that “public misunderstanding of opinion surveys can be expected to continue as long as the mass media ignore or belittle their technical intricacies,”5 this study attempts to investigate this shift.

Polls in the Media

Polls have become a ubiquitous tool for probing people’s opinions, and the reporting of public opinion polls in the news media has become a major index of public feelings about innumerable topics during the last several decades. According to a Lexis-Nexis search conducted by the authors, the New York Times and the Washington Post carried nearly 10,000 opinion poll reports during the three years of 1996, 1997 and 1998.

Even though the use of polls by the news media has been criticized as “manufactured news”6 or “pseudo-events ‘”7 the significance of public opinion polls in news reports, especially in politics or election coverage, has been well documented in the literature for decades.

Philip Meyer, in his influential 1973 book Precision Journalism, emphasized social science methodology for news reporting.8 Three years before, Harold Mendelsohn and Irving Crespi argued that the use of polls by journalists usually satisfies “the public’s curiosity about the world in which it lives.”9 Twenty years ago David Weaver and Max McCombs observed that “polling has provided journalism with a new perspective on news, and it has made journalists more aware of the news value of quantitative data.”10 Albert Gollin noted that the reporting of opinion polls by news media offers media organizations some opportunities for “altering and extending” traditional modes of news reports, especially in political news coverage.11 He argued that if the news media have become a kind of a Fourth Estate, then public opinion polls might be “one of their sharpest swords.”12 Similarly, Charles Roll and Albert Cantril noted that “if the press and mass media can be termed `the Fourth Estate’ of government, then the polling establishment is easily a candidate to become `the Fifth Estate.””3

Why are the news media so addicted to public opinion polls? Charles Salmon and Ted Glasser described several possible reasons for journalists’ use of public opinion polls, including a) “the aura of science and the prestige of scientific measurement,” b) “the [transformation of] the subjectivity of opinion into the objectivity of fact,” and c) the evasion of “the difficult question of the quality of public opinion.”14

In addition to the frequency of public opinion polls in news coverage, researchers have long studied the quality of poll reports, especially regarding the conformity to disclosure standards of the American Association for Public Opinion Research (hereafter, AAPOR).11 The AAPOR first adapted disclosure standards for reporting methodological information in 1969, including sample size, sampling error, sponsor, question wording, method of interviewing, definition of population, timing and sometimes completion rate.

Based on their findings that much reporting of poll results did not conform to the standards,16 many scholars have argued that journalists need to pay more attention to these AAPOR disclosure standards in reporting on poll results. For example, Rollberg et al. wrote: “In fact, if newspapers’ stories routinely included all eight disclosure standards, along with clearly understandable definitions of those standards, they would be serving two of the purposes of journalism: reporting the news and educating the readers.”17

Paletz and his colleagues examined the poll reports of The New York Times and the CBS and NBC evening news programs during the years of 1973, 1975 and 1977.18 They found that the credibility of polls was related to the identification of poll conductors. Their comparison of poll reports between newspapers and TV also revealed that the poll stories of The New York Times were better at reporting methodological information (e.g., sample size, timing of survey, question wording, etc.) than those of CBS and NBC.”

In other words, their findings supported to some extent the earlier argument made by Wheeler that “television is a very bad medium for the reporting of polls because it demands emphasis on the dramatic and the visually interesting at the expense of things which are subtle and ambiguous. For a poll to get on the air, it must be timely and newsworthy. It need not be correct.”20

Based on a study of three daily newspapers-the Atlanta Constitution, Chicago Tribune and Los Angeles Times-Miller and Hurd found that the number of poll reports during 1972 to 1979 had dramatically increased, but not the level of conformity to the AAPOR standards.21 They also reported that the level of conformity in election-related and press-conducted polls was higher than among non-election or externally conducted polls.22 They concluded that “newspaper editors do not see that the methodological details required for conformity to AAPOR reporting standards are automatically worthy of highly valued news space.”23 Based on an analysis of the poll reports in two statewide newspapers in Detroit, Salwen also suggested that newspapers’ own in-house polls were better than wire-service and syndicated polls in conforming to the AAPOR standards.24

Rollberg et al. examined how six U.S. daily newspapers reported public opinion polls during the 1988 presidential campaign period.25 They found that the average level of conformity to the AAPOR standards in poll reports was 36 percent, ranging from 81 percent (poll sponsor) to only 6 percent for response rate. Polls sponsored by the newspapers or wire services were more likely to conform to the AAPOR standards than were the polls from syndicated or other sources. Two statewide newspapers-the Chicago Tribune and Atlanta Constitution-did a better job of reporting methodological information than did the national newspapers USA Today and The Washington Post.26

On the other hand, in a responding essay to Rollberg et al., Meyer and Jurgensen argued that “literal conformity of every report about a poll to AAPOR standards would place a burden on the readers that would reduce, not increase, their net acquisition of information.”27 They argued that the AAPOR standards were adopted to prevent concealment of facts by survey researchers and that “a reporter should always know the eight items of information for every poll that he or she writes about,” but that doesn’t mean that all should be included in any particular news story. Rather, the decision of what to include should be based on the usual journalistic standards of interest and relevance.

Our study does not address the relationship between conformity to AAPOR disclosure standards and readers’ perceptions. As mentioned earlier, along with an examination of the subjects, content and format of poll reports, this study seeks to compare its findings with those of previous studies, especially those assessing conformity to AAPOR standards in poll reporting.

Why Online Polls Matter

Along with the increasing use of traditional polls in news media, there have been more online polls to take the pulse of the public through the Internet. According to our Lexis-Nexis search, the number of news stories reporting online polls has rapidly increased from a mere five in 1992 to 71 in 1996 to 415 in 1998.28 Given this dramatic increase, it is important to examine the reporting of online polls as compared with more traditional telephone and mail surveys.

Advocates of online polling often emphasize the relative cost and time advantages over traditional polls. Based on the Harris Internet poll’s accurate prediction of the winners in 21 of the 22 races in the 1998 gubernatorial and senatorial elections, researchers from Harris Interactive (formerly Harris Black Ltd.) argued that “Internet polling is less expensive and faster and offers higher response rates than telephone surveys.”29 Gorden Black, Harris Interactive chairman, even said “if we can do what we want to do in 2000, it will all but eliminate telephone polling going forward.”30

However, many have severely criticized online polls.31 Despite the huge number of participants, the respondents of online polls are almost always not representative of the general public. The demographics of Internet users in online polls are usually not the same as those of the general population unless a random sample of the public has been recruited and provided with access to the Web, which is rarely done. Regarding methodological limitations of online polls, Alan Rosenblatt has argued that representative online sampling still calls for a complete sample frame and reliable sampling procedures.32 Many national surveys have reported that the Internet users still tend to be too white, educated and rich to draw inferences to the overall population.33 Thus, it is usually not possible to assume that online polls are accurate measures of what people in general think.34

Regarding the reporting of online polls in the media, there have been few studies.35 Based on their content analysis of online poll reports between 1992 and 1995, Wu and Weaver found that the number of online poll stories had notably increased during these years. The topics of online polls ranged from politics and economics to entertainment and sport issues. They suggested that online poll stories were less likely to conform to AAPOR disclosure standards than traditional poll stories, but this speculation was not based on a systematic comparison of the two kinds of news reports.36

In an experimental study of the credibility of traditional and online poll reports, Kim et al. found that the trustworthiness of poll results, the honesty of pollsters, and the believability of the poll stories were all rated somewhat higher by those who read the traditional random sample poll report than by those who read the online poll version.37 Based on these findings, they suggested that news media should be careful about what kind of poll results they report and how they report them, if they want to preserve their own credibility as well as that of random sample polls. This concern with the credibility of news media as well as polls is another reason for studying how well online polls conform to the AAPOR disclosure standards .38

Research Questions and Hypotheses

As mentioned earlier, the purpose of this study is to compare how traditional (mostly, telephone or mail survey) and online polls (mostly Web site surveys and a few e-mail polls) were reported in selected U.S. newspapers during the same time period. Unlike many previous studies of political or election polls reported in newspapers,39 this present study attempts to access all kinds of opinion poll reports, regardless of their subject. In addition, this study compares the reporting of traditional polls with that of online polls. More specifically, we seek to compare the subject and sponsors of traditional and online polls and the degree of conformity to the AAPOR standards for reporting polls.

RQ1:

Is there a difference in poll conductors and subjects between traditional and online poll reports?

RQ2.

What types of sampling methods are used in collecting data for traditional opinion polls?

In addition to above research questions, the following three hypotheses regarding AAPOR standards in poll reports are proposed, based on the findings of previous research, especially Wu and Weaver’ (Hi), Miller and Hurd 41 (H2 and H3), Salwen42 (H2 and H3) and Rollberg et al.43 (H3).

H1:

The news reports of traditional polls will be more likely to conform to

AAPOR standards than the reports of online polls.

H2:

The news reports of political or election polls will be more likely to conform to AAPOR standards than those of non-political polls, especially in the case of traditional poll reports.

H3:

Polls conducted by media organizations will be more likely to conform to AAPOR standards than polls conducted by non-media organizations, especially in the case of traditional poll reports.

Methods

To answer the question of how public opinion polls are reported by the news media, we content-analyzed news media coverage of traditional and online polls during 1996 to 1998. We used two methods to create the sample of traditional and online poll news stories. Our goals were to keep the number of poll stories manageable, to be able to compare the reporting of online polls with traditional polls and to be able to compare our findings with those of previous studies of poll reporting.

The sample of traditional poll reporting was drawn from The New York Times and The Washington Post indexes in the Lexis-Nexis database. More specifically, by using systematic sampling, 156 stories were randomly selected from 9,618 poll reports (news stories, sidebars, columns, editorials, etc.) published from January 1, 1996, to December 31, 1998, in the two newspapers.

The sample of online polls consisted of all the online poll stories retrieved from the newspaper index (“PAPERS”) in the Lexis-Nexis database.44 With the key words “online poll” or “Internet poll,” the Lexis-Nexis search yielded 371 poll reports during the same time period. Based on a systematic sampling method, 185 newspaper online poll stories were randomly selected.

The poll stories selected for this study were either exclusively devoted to the reporting of poll results or used poll results as a primary part of news stories. Excluded were news items that did not include any poll data or used the term “poll” as a part of the text without focusing on the poll results. Each story was coded for year of publication, conductor of poll (e.g., media organizations, commercial polling services, research/professional institutes, etc.), focus of poll (e.g., politics, sports, life styles, entertainment, etc.), type of sampling method (e.g., nationwide random, statewide random, convenience sample, exit poll, Web site poll, e-mail poll, etc.) and conformity to AAPOR disclosure standards.

The American Association for Public Opinion Research recommends disclosure of eight categories of information to assure confidence in reports of poll and survey results:45

1. Sample size – the total number of people questioned;

2. Sponsor – the organization that had the survey done;

3. Question wording – complete wording of all the questions on which results were reported in the news item;

4. Sampling error – percentage of error at some statistical level of confidence, that is, reliability of results at plus or minus a certain percentage;

5. Definition of population – information about who’s sampled (e.g., registered voters, adults 20 or older, etc.);

6. Method of survey – method of contacting respondents to obtain interviews (e.g., phone, mail, interview, etc.);

7. Timing of survey – specific dates when the poll or survey was conducted;

8. Completion rate or number of respondents – proportion of respondents contacted who actually responded– where applicable.

About 10 percent of the total sample (35 stories, 16 for traditional and 19 for online polls) were analyzed by two coders to test intercoder reliability. Krippendorff’s Alpha was .92 for subjects, .92 for poll sponsor, and .90 for conformity to the nine AAPOR standards in general. The overall coefficient for all measures was .91, indicating a high level of coding reliability.

Findings

Poll subjects, conductors and types of sample

Table 1 shows that both traditional and online poll reports carried more non-political poll results than political/election-related poll results during the 1996-98 time period. This pattern is more evident for online poll reports. As shown in Table 1, a slight majority of traditional poll reports were about nonpolitical polls (59 percent), whereas an overwhelming majority of all the online poll reports were about non-political polls (87.6 percent).

A comparison of the specific subjects of traditional and online poll reports reveals that online poll reports were much more likely to focus on sports and entertainment than were news stories based on traditional surveys. As shown in Table 1, news reports based on traditional polls were much more likely to focus on politics and election races, and somewhat more likely to stress health/ medical issues, than were online poll news stories. The most common subjects of online poll reports were sports, entertainment, political issues and life styles.

Table 2 reveals that about 62 percent of traditional polls were conducted by non-media organizations, such as research groups and commercial polling services, whereas only 28 percent of online polls were. Clearly media organizations were the most common conductors of online polls during the 1996-98 period, mainly for sports and entertainment news. More specifically, after media organizations the most frequent conductors of online polls were commercial polling services (9.3 percent) and business or marketing services (7.8 percent), as Table 2 indicates. The most common conductors of traditional surveys, after media organizations, were research groups (18.9 percent), interest groups and community organizations (19.7 percent), helping to explain why traditional poll stories were more likely to focus on politics and health issues than were online poll stories.

In addition, this study examined the type of sampling method in traditional opinion polls (RQ2). About 27 percent of all traditional poll reports included some form of random sampling, but almost one-half (45 percent) did not include any information about type of sampling (nationwide or regional, random, convenience, exit poll and others). Thus of all traditional poll reports that did include information about sampling, 46 percent specified some form of random sampling.

On the other hand, because of “self-selected” respondents, most online poll reports did not include any information about type of sampling. Instead, in terms of polling method, about 9 percent of online poll reports included specific information about how these surveys were conducted. Of all online poll reports including information about method of survey, about 90 percent used Web sites; whereas, only 10 percent of online polls were conducted via e-mail.

AAPOR disclosure standards

We turn now to the question of how well news reports of traditional and online polls conformed to the standards of disclosure recommended by the American and World Associations for Public Opinion Research.

The first hypothesis of this study predicted that the news reports of traditional poll results would be more likely to conform to AAPOR standards than would the reports of online polls. The figures in Table 3 indicate considerable support for this prediction, with significantly more stories based on traditional polls including information about timing, sample size, definition of population, method of interviewing and sampling error. There was no significant difference between the two types of poll stories on disclosure of the sponsor of the survey or the question wording, and news reports of online polls were more likely to mention number of respondents than those based on traditional surveys. But, overall, Hypothesis 1 was supported for five of the eight standards where there were any data available. None of the news reports included response rate, which is a noteworthy finding in itself.

Table 3 also provides a comparison of the level of conformity to AAPOR standards between the years examined – the late 1980s (1988) of the Rollberg et al. study and the late 1990s (1996-1998) of this study. Specifically, for six of the eight standards in news reports of traditional polls, the conformity levels with the AAPOR disclosure standards decreased, especially for sample size and definition of population. The two standards showing some slight improvement over the decade were sampling error and question wording. The average percentage of compliance also decreased slightly from 36 percent to 29 percent, but Rollberg et al. studied six newspapers (one of which was the Washington Post) compared to only two in our study, so this conclusion is only tentative in the absence of a tighter comparison.

Overall, contrary to the findings of previous studies in which the improvement in the reporting of methodological information in poll stories was found over the time during the 1970s and 1980s,’ this study suggests that the overall conformity level with AAPOR standards likely decreased somewhat in the late 1990s, compared to the 1980s.

The second hypothesis predicted that the news reports of political or election polls would be more likely to conform to AAPOR standards than would those of non-political polls, especially in the case of traditional poll reports.

Table 4 indicates virtually no support for this prediction. Although statistical differences were not found, of the eight standards in the news reports of traditional polls, five were in the opposite direction-with the news stories of non-political polls being more likely to mention the sponsor, sample size, definition of population, method of survey and question wording than were the reports of political polls. Only three standards (timing, sampling error and number of respondents) supported Hypothesis 2. Political poll reports were more likely to carry information about sampling error than the reports of nonpolitical polls, probably because of the emphasis in many political polls on the gap between competing candidates in elections.

Regarding online poll reports, only one variable (method of survey) supported the prediction at the .05 significance level. Overall, there were no statistically significant differences in conformity to the disclosure standards in the news stories about online polls, although there was a slight tendency for the stories about political online surveys to be more likely to mention sponsor, definition of population, number of respondents and question wording than those about non-political polls, as predicted by Hypothesis 2.

The third hypothesis predicted that polls conducted by media organizations would be more likely to conform to AAPOR standards than would polls conducted by non-media organizations, especially in the case of traditional poll reports.

Overall, Table 5 reveals that this prediction was not supported. Although only two measures (sample size and definition of population) were significantly different between media and non-media polls in news reports of traditional surveys, these and three others of the eight standards were in the opposite direction predicted-with non-media poll reports being more likely to include timing, sample size, definition of population, method of interviewing and number of respondents than news stories based on media-conducted polls. The reporting of sampling error and question wording was equally likely in both kinds of traditional polls.

Discussion and Conclusions

This study suggests several conclusions. First, the reporting of online polls in selected U.S. newspapers in the late 1990s was much more likely to focus on non-political subjects such as sports and entertainment than were the news reports of traditional surveys. This could be considered somewhat comforting to those concerned about the accuracy of the political information being conveyed to citizens by news reports of online poll results, especially polls conducted during hotly contested election campaigns.

Second, this study suggests that the reporting of traditional polls in the two prominent newspapers was more likely to be based on opinion polls conducted by non-media organizations, such as research groups, educational institutions and commercial polling services, than polls conducted by their own or other media organizations. Online poll reports, on the other hand, were more likely to be based on surveys conducted by media organizations. This could be considered a cause for concern. Because most online polls were being conducted by media organizations, the results of surveys conducted by various interest groups, community and other organizations were not reported nearly as often as those conducted by the news media themselves.

As Gollin put it 20 years ago, “news creation by polls offers the press some opportunities for altering and extending traditional modes of coverage of politics. But it poses some dangers as well, which deserve close attention.”47

Third, the findings suggest that traditional survey articles carried in newspapers are much more likely to conform to the AAPOR disclosure standards than are online poll reports, but even these newspaper articles don’t usually disclose much information about the surveys, except for the sponsor. This study also found that the reporting of most AAPOR standards in traditional poll reports probably decreased from 1988 to the late 1990s.

Clearly, the reporting of online opinion surveys did not include nearly as much information about the conduct of these polls as did the reporting of traditional polls, except for the sponsor, the question wording and the number of respondents. Important information about timing, definition of population, sample size, interviewing method and sampling error was almost nonexistent.

Given the lack of methodological information in poll reports and the increasing reporting of online polls, however, this could be another cause for even more concern among those who care about improving public understanding of survey research and public opinion, especially public opinion about political issues.

As more online opinion polls are reported in traditional news media, such as newspapers and television, it may become more difficult for people to discriminate between the two types of polling. Moreover, considering the increasing use of online opinion polls by the news media and their effects on people’s beliefs and attitudes, the questions of what kind of poll results journalists report and how they report them are very important. Presumably non-representative online polls mostly conducted by media organizations themselves can provide people with inaccurate information of what the general public thinks about a certain issue. This is why the reporting of online polls by news media matters.

Finally, this study has some limitations. This paper aimed to keep the number of online poll stories manageable to be able to compare the reporting of online polls with traditional polls. Thus, this study compared reporting of traditional opinion polls in two elite national newspapers with the reporting of online polls in many different newspapers. This means that any differences could be partly a function of the different newspapers and not necessarily of differences between traditional versus online poll reports. However, the large differences we found in this study suggest that not all the variation can be explained by the difference in newspapers, given the similarities in reporting AAPOR standards by different newspapers found in earlier studies.

Second, the lack of support for the second and third hypotheses of this study is puzzling. Given that more online opinion polls are reported in news media, we hope that future research based on more representative and comprehensive data than ours will test these hypotheses.

Notes

1. Kathleen A. Frankovic, “How Polling Becomes News: Communicating the Counting of Public Opinion,” Political Communication, Special Issue: MultiMedia Politics (CD-Rom, 1998).

2. See, for example, Herbert Asher, Polling and the Public: What Every Citizen Should Know, 2nd ed. (Washington, D.C.: CQ Press, 1992); S. R. Gawiser and G. E. Witt, A Journalist’s Guide to Public Opinion Polls (Westport, Conn.: Praeger, 1994); Sung Tae Kim, David H. Weaver, and Lars Willnat, “The Reporting and Perception of Online Polls and Message Credibility” (paper presented at the Annual Conference of the Midwest Association for Public Opinion Research (MAPOR), Chicago, Ill., November 19-20, 1999); Alan Rosenblatt, “On-Line Polling: Methodological Limitations and Implications for Electronic Democracy,” The Harvard International Journal of Press/Politics 4, no. 2 (spring 1999): 30-44; Michael W. Traugott and Paul J. Lavrakas, The Voter’s Guide to Election Polls (Chatham, N.J.: Chatham House Publishers, 1996); Lars Willnat, David H. Weaver, Sung Tae Kim, and Brad McKay, “Perceptions of Traditional and Online Polls and the Third-Person Effect: How Americans Think about Polls and Their Effects” (paper presented at the annual meeting of IAMCR, Leipzig, Germany, 1999); Wei Wu and David Weaver, “On-Line Democracy or On-Line Demagoguery? Public Opinion Polls on the Internet,” The Harvard International Journal of Press/Politics 2, no. 4 (fall 1997): 71-86.

3. Kim, Weaver and Willnat, “Online Polls and Message Credibility;” Wu and Weaver, “Public Opinion Polls.”

4. Jack lN. Germond, “The Impact of Polling on Journalism,” in Polling on the Issues, A. H. Cantril, ed. (Cabin John, Md.: Seven Locks Press, 1980); Philip Meyer, Precision Journalism: A Reporter’s Introduction to Social Science Methods (Bloomington, Ind.: University Press, 1973); David H. Weaver and Maxwell E. McCombs, “Journalism and Social Science: A New Relationship?” Public Opinion Quarterly 44, no. 4 (winter 1980): 477-94.

5. Leo Bogart, Silent Politics (New York: Wiley-Interscience, 1972), 23.

6. Nicholas Von Hoffman, “Public Opinion Polls: Newspapers Making Their Own News?” Public Opinion Quarterly 44, no. 4 (winter 1980): 572-73.

7. Albert E. Gollin, “Exploring the Liaison Between Polling and the Press,” Public Opinion Quarterly 44, no. 4 (winter 1980): 445-61.

8. Meyer, Precision Journalism.

9. Harold Mendelsohn and Irving Crespi, Polls, Television and the New Politics (Scranton, Pa.: Chandler Publishing Co., 1970),13.

10. Weaver and McCombs, “Journalism and Social Science,” 478.

11. Gollin, “Exploring the Liaison.”

12. Gollin, “Exploring the Liaison,” 446.

13. Charles W. Roll, Jr. and Albert H. Cantril, Polls: Their Use and Misuse in Politics (New York: Basic Books, Inc., 1972),11-12.

14. Charles T. Salmon and Theodore L. Glasser, “The Politics of Polling and the Limits of Consent,” in Public Opinion and the Communication of Consent, eds. T. L. Glasser and C. T. Salmon (New York: The Guilford Press, 1995), 443-44.

15. See, for example, M. Mark Miller and Robert Hurd, “Conformity to AAPOR Standards in Newspaper Reporting of Public Opinion Polls,” Public Opinion Quarterly 46, no. 2 (summer 1982): 243-249; David L. Paletz, et al., “Polling in the Media: Content, Credibility, and Consequences,” Public Opinion Quarterly 44, no. 4 (winter 1980): 495-513; J. N. Rollberg, L. W. Sanders and M.D. Buffalo, “Down to the Wire: How Six Newspapers Reported Public Opinion Polls during the 1988 Presidential Campaign,” Newspaper Research Journal 11, no. 4 (fall 1990): 80-92; Michael B. Salwen, “The Reporting of Public Opinion Polls During Presidential Years, 1968-1984,” Journalism Quarterly 62, no. 1 (spring 1985): 272-77; Wu and Weaver, “Public Opinion Polls.”

16. See, for example, Miller and Hurd, “Conformity to AAPOR Standards;” Rollberg, Sanders, and Buffalo, “Down to the Wire;” Salwen, “Reporting of Public Opinion Polls.”

17. Rollberg, Sanders, and Buffalo, “Down to the Wire,” 91.

18. Paletz, et al., “Polling in the Media.”

19. Ibid.

20. Michael Wheeler, Lies, Damn Lies, and Statistics (New York: Liveright, 1976), 202.

21. Miller and Hurd, “Conformity to AAPOR Standards.”

22. Ibid.

23. Miller and Hurd, “Conformity to AAPOR Standards,” 248.

24. Salwen, “Reporting of Public Opinion Polls.”

25. Rollberg, Sanders, and Buffalo, “Down to the Wire.”

26. Ibid.

27. Philip Meyer and Karen Jurgensen, “Beating Disclosure to Death: A Rejoinder to Rollberg, Sanders and Buffalo,” Newspaper Research Journal 12, no. 3 (summer 1992): 3, 6.

28. We conducted a search on all stories in Lexis-Nexis with the key words “online poll.” Included were the following six categories: magazines (MAGS), major newspapers (MAJPAP), newspapers (PAPERS), transcripts (SCRIPT), wire services (WIRES) and newsletters (NWLTRS).

29. John Simons, “Are Political Polls via Internet Reliable? Yes? No? Maybe?” Wall Street Journal Interactive Edition, (13 April 1999).

30. Ibid.

31. See, for example, Rosenblatt, “On-Line Polling;” Willnat, “Perceptions of Traditional and Online Polls;” Wu and Weaver, “Public Opinion Polls.”

32. Rosenblatt, “On-Line Polling.”

33. See, for example, S. Ross and D. Middleberg, “The Media in Cyberspace Study II,” Tools for Success, 1996, (20 July 2001); Emerging Technologies Research Group, “American Internet User Survey,” 12 May 1995, (20 July 2001).

34. Richard Morin, “Unconventional Wisdom: New Facts and Hot Stats from the Social Sciences,” The Washington Post, 27 December 1998, sec. C5.

35. See, for example, Kim, Weaver, and Willnat, “Online Polls and Message Credibility;” Wu and Weaver, “Public Opinion Polls.”

36. Wu and Weaver, “Public Opinion Polls.”

37. Kim, Weaver, and Willnat, “Online Polls and Message Credibility.”

38. Ibid.

39. See, for example, Miller and Hurd, “Conformity to AAPOR Standards;” Rollberg, Sanders, and Buffalo, “Down to the Wire.”

40. Wu and Weaver, “Public Opinion Polls.”

41. Miller and Hurd, “Conformity to AAPOR Standards.”

42. Salwen, “Reporting of Public Opinion Polls.”

43. Rollberg, Sanders, and Buffalo, “Down to the Wire.”

44. “PAPERS” is a combined file of full-text news sources in the Lexis-Nexis database for more than 250 daily or weekly U.S. newspapers, including the most prestigious such as The New York Times and The Washington Post. Lexis-Nexis, Lexis-Nexis, .

45. “Code of Professional Ethics and Practices,” American Association for Public Opinion Research, 1994. Photocopy provided by AAPOR. For more detailed information about the AAPOR standards, see G. C. Wilhoit and D. H. Weaver, Newsroom Guide to Polls & Surveys (Bloomington, Ind.: Indiana University Press, 1980), 78-79.

46. See, for example, Miller and Hurd, “Conformity to AAPOR Standards;” Salwen, “Reporting of Public Opinion Polls.”

47. Gollin, “Exploring the Liaison,” 445.

Kim is a doctoral candidate and a Roy W. Howard Research Assistant, and Weaver is the Roy W. Howard Research Professor in the School of Journalism at Indiana

University. They gratefully acknowledge the support of the Roy W. Howard Chair for this research and the constructive suggestions of NRJ’s senior research editor Guido H. Stempel III, distinguished professor emeritus, Ohio University.

Copyright Newspaper Research Journal, Department of Journalism, University of Memphis Summer 2001

Provided by ProQuest Information and Learning Company. All rights Reserved