Mi? PERCEPTION OF BIAS IN NEWSPAPERS IN THE 1 6 ELECTION Bean Baker * Charles Cannell University of Michigan In the past several national political campaigns there have been"maaerenen complaints, particularly from Democratic politicians, that the newspapers 1 handling of campaign news was biased. Supporters of Franklin D, Roosevelt often claimed that a majority of the newspapers opposed the New Deal. More recently both Harry Truman and Adlai Stevenson made critical comments about what they termed the "one-party press" While the accusation originally referred only to editorial support, I t has since come to have a more general context and to imply bias in the handling of political news. Numerous studies have been conducted in an attempt to answer the question as to whether the news is presented objectively or is slanted in favor of one or the other political party. Little research, however, has been directed to the consumer of the news to see how he feels about the political news in the newspapers. Many readers are aware of the editorial stand taken by the papers they read, and many also are aware of the accusation of news bias. Does this mean that the public feels that the newspapers are biased? This topic was the subject of the research to be reported here. Specifically, this study is directed at three questions: 1«Does the public perceive press treatment of presidential campaign news to be biased or objective? *This analysis was carried out as a class project in a course in research methods. Much of the work was done by three graduate students in the Journalism Department, Ester Celi, Culver Eis nj5eis, and Joseph Zcbian*
2 2. Does this perception differ for Democrats and Republicans? 3. Does perception of bias influence selection of the mass media relied upon as a source of campaign news? The data were gathered in two surveys, one a national crosssection study conducted by the Survey Research Center of the University of Michigan and the other conducted in the city of Ann Arbor. The interviews on the national study were taken shortly after the 1956 presidential election. Approximately 1800 persons were interviewed. Sdjncs:.we were limited severely in the number of questions which could be included in the national study, a small sample of Ann Arbor residents were interviewed using questions which we hoped would illuminate and amplify the findings of the larger sample. This study of 223 people was conducted as part of a class project by students in courses In research methods in the Departments of Journalism and Psychology. I t XIZTCJCJ be mentioned here that while the national study achieved a response rate of about 85$, in the Ann Arbor study only 67$ of eligible respondents were interviewed. This low response rate leads one to interpret the findings with considerable caution. The major findings in this report are based on national data. In talking about the newspapers, the respondents were not restricted to the papers which they themselves read. Rather, they were encouraged to respond to v.the newspaper" as an institution. Thus we have to a large extent a stereotypic response, although a number of respondents said they could answer the questions only about the paper with which they were most familiar. After questions on the use of the media for news about the campaign, the respondents
3 were asked: "In reporting the news about the campaign and the candidates, do you feel that the newspapers were equally fair to both sides or not?" If the respondent said he thought they were unfair, he was asked, "Which candidate did the papers favor?". Table I gives the responses to these two questions. This table and most of those which follow include only those respondents who reported reading about the campaign in the newspaper. The last column of Table I shows that approximately one-quarter of the sample thought the papers favored one of the two major political candidates or parties, and most of these perceived a pro-eisenhower slant. The others thought that the papers were equally fair to both sides or responded that they didn! t know. In one sense, those who responded "don't know" belong to the "equally fair" group, since they did not perceive favoritism. They are kept separate, however, because they are a significant group for some of the analyses which follow* Table I The first three columns of Table I show the relationship between voting behavior and the perception of bias. Slightly more than half of the Stevenson voters saw no bias in news, while about one-third saw the news as slanted to favor Eisenhower. Eisenhower voters were more pronounced in the opinion that the papers were fair to both sides, but one in six of them said they thought the papers favored their candidate. As one might expect, the opinions of persons who reported that they did not vote fell between the two parties.
h The "Don't know" response was highest for this group, possibly reflecting a general lack of sensitivity to political issues. In these findings we see confirmation of the adage that i f a statement runs counter to one's own point of view, we are more likely to consider i t biased than i f i t agrees with our stand. Even among the Republicans, however, about one out of five perceived the news as unfair and most saw i t as favoring their own candidate. One might suppose that most of those who thought the newspapers unfair would have a definite basis for their reactions, and would be able to cite evidence on how bias was demonstrated and might even have opinions on why bias existed, but this waa not found to be the case in the Ann Arbor survey. Respondents who perceived bias were asked, in an open question, in what ways newspapers were unfair. About half said they didn't know. Those who did have opinions said that the papers were unfair in the amount and in the placement of news about the candidates. Others said bias was indicated by the selection of items to be reported and that headlines and pictures were such as to favor one candidate. Since such a high proportion of respondents were unable to report a basis for their feelings that the news was presented unfairly, i t raises the question as to whether readers were really conscious of bias in the papers from personal experience or whether they were merely reporting statements they had read or heard that the press was biased. Some information later in this report will shed some light on this question.
Since the question relating to voting pertained only to the vote for the president, and because many people crossed party lines in the 1956 election to vote for Eisenhower, i t is interesting to see how perception of bias related to the respondent's political affiliation or party identification. Respondents had been asked a series of questions on party identification which permitted us to classify respondents as Republicans, Democrats, or Independents. Table I I The responses of these three groups to the answers to the questions on bias were analyzed separately. Table I I gives the breakdown for those who identify with the two major parties and those who were classified as independents. Comparing Tables I and I I, we see the effects of Democratic voters for Eisenhower. Among the hard core of Democrats who voted Democratic, 36$ perceive a Republican bias in the paper. Among nominal Democrats, many of whom voted for Sisenhower, the perception of bias drops to 32$. Among both Republican voters and Republican identifiers, however, 1?$ see a Republican bias. The data on party identification and bias was also tabulated separately for four geographic regions of the country. I t was felt that because of the larger proportion of Democratic papers in the South, the pattern might be reversed, with the southern Democrats seeing less bias and the Republicans perceiving a greater Democratic For a discussion of the methods and uses of data on party identification see The Voter Decides, by A. Campbell, 0. Gurin, W. Miller, Row, Peterson, and Company, 195U.
6 bias. The data did not bear this out. Thare was some tendency among southern respondents, especially southern Republicans, to see a greater Democratic bias; but the differences are not statistically significant. Table I I I I t was expected that education would be ai f actor in perception of bias, and this in fact proved to be the case. When respondents were separated into three groups, on the basis of educational level, i t was found that the higher the education the more likely the tendency to see the news as favoring one candidate. Respondents in the lower educational groups were less likely to perceive bias and had a greater tendency to respond with "don't know" to the question on favoritism, than did those with greater education. Table IV Since both education and party identification are found to be related to perception of bias, the question arises; how are these factors interrelated? Table IV shows these relationships«i t is clear that both education and party identification are important in perception of bias. Low education Republicans are least likely to see bias. "fair" responses, but Republicans make up a larger proportion of this proportion declines with higher education. Fewer Democrats generally see the papers as fair, and here also an increase in bias perception occurs with higher education*
7 Independents axe not clearly different from other respondents, but here also the proportion judging the papers as "fair" declines with higher education. Having established the pattern of the perception of bias as related to party identification and education, the question now arises; How does this perception relate to other attitudes and to the use of the newspaper in comparison with other media? Before being questioned about their perception of fairness of news, respondents were asked about their use of the newspapers and other media during the Presidential campaign. One can hypothesize that i f a person perceives one medium as unfair, he will depend less upon that medium than upon others far his information. To test this hypothesis, respondents were asked; "Of all ways of following the campaign, which one would you say you got the most information from - newspapers, radio, television, or magazines?" The percentages of those who saw bias and of those who did not were compared for the media they "most used" for campaign information. In Table V we see that the split between the two dominant media, newspapers and television, is not significantly different among those who perceived bias and those who did not. Table V Apparently perception of bias made no significant difference to the respondent's behavior toward the newspaper. When Democrat and Republican party identifiers were compared on the same data, the differences were slightly larger but not significantly so 0
8 The interpretation of these findings is difficult and one conjecture probably is as valid as another. As mentioned earlier, i t may be that respondents have only weak, poorly formed opinions on the topic of newspaper fairness rather than firm perceptions or strongly held attitudes. I f these feelings are vague and amorphous, they can be expected to have only slight effect upon behavior. Some evidence for this view is found in the way the questions were answered. From reading the Interviews, i t appears that few respondents volunteered comments or gave other indications of strong feeling. The high proportion of respondents who answered "don! t know" is * another Indication that the perceptions are not well formed and the opinions weak. As we have already mentioned, about half the re spondents did not reply to the query as to how favoritism was shown* We would like to have more information which would shed light on these findings. Some results from the Ann Arbor study are relevant here. First, we should say that in general the findings of the Ann Arbor study were comparable to the national data for the questions already reported. The differences were sharper in Ann Arbor. These differences reflect the fact that Ann Arbor is predom5j3antly a University town and that the educational level of the respondents was considerably above the national average. The findings for Ann Arbor generally looked like the^cclloge population for the country as a whole. Therefore, the comparisons of this survey with national data must be made with caution.
9 Ann Arbor respondents were asked, "How about accuracy? Did you get the most accurate information about the campaign from newspapers, radio, television, or magazines?" and "Which one gave the least accurate information?" These questions were asked prior to those on perceptions of fairness in the newspapers* Table VI Table VI, on perception of bias, and opinions as to which medium was most accurate and least accurate, shows a relationship. More than a third of those who saw no bias considered the newspapers the most accurate of the four media. Of those seeing bias, about half said that television was the most accurate medium, and that newspapers were least accurate. From these data we get some feeling that report of perception of bias is not an entirely superficial reaction. At least in the Ann Arbor sample we have an independent response which substantiates the validity of bias perception. No study of bias perception, i t would seem, should overlook the public's view of those who direct newspapers* Two final questions were therefore asked respondents in the Ann Arbor survey to obtain a general notion of how they perceived the policy-makers of the press. The questions read: "Who do you think decides which candidate the paper will favor? Is i t the owner, the person who writes the news, or someone else?" The other read: "In general, in every newspaper someone must decide how i t will be run, that is, what
10 news to print, how to write i t, and so forth. Who do you think usually decides such.things?" Answers were coded and are presented in Table!VEI with percentages of respondents in the sample who gave each in reply to the two questions. Table VII I t is interesting to note whom people seem to perceive as the makers of important decisions on newspapers. More than half thought the editor or editorial staff were responsible for general policy, but nearly half, replying to the question on choice of candidate to be supported, said that the owner or board of directors of a newspaper decided this 0 Of equal significance is the proportion of respondents who couldn 1 1 answer the question. From this and other impressions of reading interviews, we suspect that there is a considerable amount of ignorance regarding newspapers and how they operate. Attempts to find a correlation between perception of bias and perceptions of the policy-makers, however, failed to show any significant relationship. For one thing, responses to the policy-maker questions were widely scattered over many kinds of answers and in analysis appeared unrelated generally to the respondents 1 opinions of news fairness.
Table VI. Perception of Bias Compared to Choice of Media Considered Most and Least Accurate (Ann Arbor Survey) Most Accurate Newspapers Newspapers Fair Biased Least Accurate New sp apers Newspapers Fair Biased Newspapers 38$ 28$ 18$ 52$ Radio 8$ 10$ 10$ 12$ Television 26$ 51$ 1W 10$ Magazines 7$ 6$ 18$ 20$ Combination of Media 5$ 3$ 1$ * Don't Know 10$ 0$ 29$ 3$ No Difference W.0$ 7$ 0$ Not Ascertained 2$ 2$ 3$ 3$ loqf 100$ 100$ (n-135) (n-61) (n-135) (n-61) Table VII. Perception of Policy-Making on Newspapers Decides Choice of Candidate Decides General Policy Publisher or Manager 3$ 5$ Editor or Editorial Staff 12$ 58$ Owner or Board of Directors hl% 18$ Writers or Reporters 12$ 1$ The Public, Including Influential People W 2$ "Top Staff" or Staff Generally 1$ 3$ Business Manager 1$ Don't Know 17$ 11$ Not Ascertained m lbbt (n-223) 1$ ioit" (n-223)
PERCEPTION OF BIAS IN NEWSPAPERS IN THE 1956 El CTION Dean Baker Charles Cannell University of Michigan Table I. Newspaper Readers 1 Perception of Bias and Presidential Vote Dem. Rep. Did Not Vote Total News Equally Fair to Both Sides 56$ 7\$ 63$ 66$ News Favored Eisenhower 36 17 22 2h News Favored Stevenson 2 U 2 3 Don't Know 5 5 i l 6 Not Ascertained 1 1 2. 1 100? 100? 100$" 100? (n-u02) (n-590) (n-187) (n-ll85) Table I I. Newspaper Readers' Perception of Bias and Party Identification Dem. Ind. Rep. News Equally Fair to Both Sides 61$ 73$ 78$ News Favored Eisenhower 32 2k 17 News Favored Stevenson 3 3 h DorJt Know 1 100? 100? " (n-502) (n-282) 1 100? (n-3li9) Table I I I. Newspaper Readers' Perception of Bias and Education Grade Sch. High Sch. College News Equally Fair to Both Sides 68$ 67$ 63* News Favored Eisenhower 19 23 29 News Favored Stevenson 3 3 a Don't Know 9 6 3 Not Ascertained 1 1 1 100$ 100$ 10C# (n-260) (n-61i7) (n-3q3)
Table IV* Newspaper Readers' Perception of Bias, Party Identification and Education Grade Sch. High Sch. College News Equally Fair to Both Sides News Favored Eisenhower News Favored Stevenson News Equally Fair to Both Sides News Favored Eisenhower News Favored Stevenson News Equally Fair to Both Sides News Favored Eisenhower News Favored Stevenson 69% 28 3 loo? (n-99) 72$ 26 2 100$ (n-u7) 6% 32 3 ICO? (n-285) 77$ 20 3 loo? (n-l61i) 61$ 35 k 100$ (n-110) 63$ 33 a 100$ (n-7d m 79$ 72$ 10 16 25 6 5 3 locf 100$ 100% (n-8u) (n-152) (n-109) Table V. Perception of Bias Compared to Most Used for Information Choice of Media Media Chosen "Most Used" Newspapers Fair Newspapers Biased Newspapers 37$ 35$ Radio 7 k Television U7 5o Magazines 5 7 Combination of Media U h 100$ 100$ (n-799) (n-323)