Association for Educational in Journalism and Mass Communication (AEJMC) Annual Conference, August 1989, Washington DC

In recent years presidential elections have been monitored by scholars and research organizations to determine if there is any trace of bias in the coverage of the campaign by the press. Bias, or preferential treatment, can be evaluated in a number of ways. Some studies look at the amount of space allocated to each candidate, others consider treatment such as position on the page, and some have evaluated the negative, positive, and neutral content.

In a pioneering study using assertion analysis Westley, et. al., investigated statements from the l960 election and found partisan patterns in the way newsmagazines covered the candidates. Stempel found that George Wallace was clearly treated as a minor candidate in the l968 election, receiving slightly more than half as much space as either of his opponents. Evarts and Stempel, however, found no identifiable bias in the coverage of the 1972 election by the networks, newsmagazines, and major newspapers.

More recently, the Center for Media and Public Affairs, a nonpartisan study group, evaluated network broadcast coverage of candidates during the primaries in the 1988 election. This study looked at positive or negative statements relating to character, job performance, campaign performance, and issue stands and found that the tone of the coverage was generally positive before the New Hampshire primary and turned negative after that point. The study found that Robertson suffered the most, while Jackson suffered the least from negative coverage.

This study focuses on the question of preferential treatment by looking specifically at the visual coverage of the candidates in U.S. newsmagazines. It is a replication--and a continuation--of a study of the 1984 campaign previously published in Journalism Quarterly.

Visual Communication Theories

One area of particular interest is the impact of visual communication in the highly image-oriented political arena. Is there any partisanship apparent in the way visuals are used to chronicle the campaign and depict the candidates? Literature discussing this issue is found in such theoretical areas as visual communication, information processing, person perception, agenda setting, and video and film aesthetics.

Visuals are often discounted in the word-oriented journalism environment, however, researchers in visual communication have long argued that pictures in mass communication messages convey important information that is attended to, processed, and remembered long after the words are forgotten. Visuals are more than decoration; they perform important roles in communication such as conveying realism, credibility, and attitudes. As Roy Blackwood noted in a discussion of realism and its effect on credibility, "In some cases the photos are the only representation of an event. Visuals are important as conveyers of information and shapers of attitudes."

In a landmark study of the information processing of news, Doris Graber found that news pictures are invaluable for forming opinions about people, including political leaders, and that they are used to appraise credibility, attract and hold the viewer's attention, stir emotions, and produce positive and negative feelings.In terms of learning she found that viewers believed that the pictures allowed them to form more complete and accurate impressions of people and events. She concluded that viewers in her studies confirmed what the visual communication literature suggests, namely that pictures make information transmission more realistic, accurate, and touching than is possible in purely verbal messages.

Grabner also found that different types of visuals command different patterns of attention. Closeups of familiar people, for example, were most likely to be noted and processed. Unfamiliar people who become noteworthy to the audience because they are in a position to express their views are also attention getting. Pictures that provide factual information or that further clarify verbal information, however, are largely ignored, as are establishing shots and distance shots.

The person-perception literature focuses on self-presentation theories and techniques. When candidates use television to project themselves to voters, for example, they are primarily engaged in a form of what Kaid and Davidson call "pseudo-interpersonal communication" in which they use television's visual intimacy to portray themselves in certain ways. The presentation of a candidate's self through media is similar to an actor playing a political role and, as Nimmo and Savage, explain, the role playing projects a charactertisic style that reflects how the candidate wants his or her performance and personal qualities to be perceived.

Expression, gestures, posture, setting or environment, and interaction with others are all tools used by people when they assume a role. Graber found in her studies that pictures of human beings were rich information sources--facial coseups, body cues including movements, posture, or grooming--and that these cues disclosed among other things a person's physical well being, poise and vigor, economic status, and sense of conformity. She also found that most people have learned to draw inferences from physical appearance and movements and that this is particularly valuable at election time when judgments need to be made about a candidate's character, trustworthiness, mental acuity, and physical vigor.

While image is partially under control of the candidate in terms of self-presentation, the "re-presentation" of the image can be manipulated by the medium through its gatekeeper role. Editorial decisions about which pictures to use affect how a candidate's image is presented and perceived. This treatment can reflect whether the candidate is treated "presidentially" or not seriously. The "play" of the visual also affects reader perception. While editing is presumed to be objective in the U.S. press, it is easy to see where subtle messages can be conveyed in the way visuals are presented.

Ann Marie Barry discovered in an analysis of the CBS Reports documentary "Teddy," that the re-presentation of an event in the media can be manipulated through the selection of content, framing, and editing of visuals. Size, for example, is an extremely important cue. An agenda-setting experiment by Wanta found that the size of photos in newspapers can have an immediate influence on readers. The study found that editors "have the power to raise their reader's salience on certain issues over a short period of time by merely increasing the size of photographs."

The visual aesthetics literature is derived from studies of photography, cinema, and video. The language of film has been codified and the nuances of meaning conveyed by photographic conventions such as setting, lighting, framing, and camera angle have been analyzed extensively. A camera angle from on high, for example, is used to "look down" on a subject and demphasize importance, while a camera angle from below is used to create a monumental image.

The Original Study

As mentioned earlier, the question of preferential treatment in visual coverage of the campaign was investigated for the first time in a study of newsmagazine photographs during the 1984 campaign. The research questions guiding that study--as well as this current study of the 1988 campaign--ask if both candidates and parties are given equivalent amounts of space and position, or does one candidate receive more emphasis? Furthermore are subtle visual communication techniques being used to communicate either negative or positive images?

The original study found that President Reagan, the incumbent, received significantly more favorable play in visuals than did the challenger, Walter Mondale. In terms of sheer quantity, Reagan photos outnumbered Mondale photos significantly. However, while Reagan was presented more favorably early in the campaign, this was reversed by the end of the campaign.

Likewise, early in the campaign Bush was presented more favorably as a vice-presidential candidate than was Ferraro; however, that changed as the amount of Ferraro coverage overtook Bush's coverage later in the campaign. Overall Ferraro photos significantly outnumbered Bush photos undoubtedly representing her news value as the first female candidate in a presidential campaign. For both presidential and vice-presidential races, the democratic candidates received more favorable treatment at the end, possibly a reflection of the underdog or come-from-behind position of the candidates. The data, however, indicate that overall the candidates were differentially represented in the newsmagazines' visuals during the l984 campaign.

This content analysis generally replicated the previous study of newsmagazine photographs. It examined all the visuals--both photographs and illustrations--of the 1988 presidential and vice-presidential candidates printed in the three national weekly newsmagazines--U.S. News & World Report, Time, and Newsweek--between the labor day kickoff and a week after the general election (September 5 to November 21).Every photo containing a candidate picture was included in the study including those on the cover and the contents page. Those photographs containing the candidates were coded twice--once for each candidate--and treated as two separate photographs in the analysis.

Coding Procedure

Coding difficulty is one of the reasons noted in the literature for the lack of attention to visual communication, although scholars noting the problem have called for more work in this area. This study attempts to construct a coding procedure that looks at preferential treatment in terms of size, position, and various dimensions of visual language that can be manipulated to cue a more positive or negative interpretation. Each photo was coded to identify candidate, magazine, publication date, size (in picas), and 15 visual attributes. The visual attribute -included activity, posture, arm position, hands, eyes, expression, interaction, camera angle, portrayal, position, size, props, setting, dress, and family association.

On the coding sheet, 11 of the visual attributes were rated more favorable (+1), less favorable (-1), or neutral based upon a set of guidelines which were developed from evaluation criteria found in the literature review. (For ease of tabulation and analysis, these values were changed to 3, 2, and 1.) (See Table 1 for operational definitions.)

________________________________________

________________________________________

Index Construction. Three indices were constructed from the sets of variables. The "behavior" index was a total of activity, posture, arms, hands, eyes, and expression. The "context" index included props, setting, dress, and interaction with colleagues and family. The "perspective" index included the position, size, camera angle, and portrayal.

Presidential and vice-presidential attributes were tested by means of two-tailed T-Tests, and ANOVA techniques were used to determine if the candidates, newsmagazines, or the time period in the campaign interacted with the size of the pictures published by the newsmagazines.

Intercoder Reliability. The two principal researchers double-coded the November 21 issue of Newsweek in order to develop an estimate of intercoder reliability. The Holsti method gave a coefficient of reliability of 84% for the 28 photos (or 448 judgements) evaluated in the test magazine This compares favorably with Graber's work which she describes as having a high reliability of 89% in its coding of visuals.

There were differences in the number of visuals by candidate and by party; however, the pattern of newsmagazine coverage for the four candidates did not vary significantly (see Table 1). From a party standpoint, republican candidates accounted for 164 visuals, while democrats accounted for 120 visuals, and the difference in visuals between the two parties was significant (X2 = 6.82, df=1, p<.01).

Overall, the four candidates appeared in 284 visuals during the test period. George Bush appeared in 125 visuals; Dukakis, 107; Quayle, 39; and Bentsen, 13. The differences between all four of these totals varied significantly from what would have been a chance distribution of photos (X2 = 121.12, df=3, p<.01) with the result that both presidential candidates appeared in significantly more photos than might have occurred by chance, while the vice-presidential candidates appeared in significantly less photos.

Time magazine published 108 visuals; U.S. News & World Report, 96; and Newsweek carried 80; but the overall differences between the newsmagazines were not significant (X2 = 4.16, df=2, p>.05). Besides analyzing the number of candidate photos in each newsmagazine, a crosstabulation was conducted to test the pattern of coverage among the three newsmagazines during the test period. In the presidential race George Bush was in 44% of the visuals published by all three newsmagazines, which was more than Michael Dukakis who appeared in 37%. In the vice-presidential race Dan Quayle was in 14% of the visuals, and Lloyd Bentsen appeared in a mere 5%.

Of the 284 total, 38% appeared in Time, 34% appeared in U.S. News & World Report, and 28% appeared in Newsweek. Crosstabulation statistics did not reveal significance in the pattern of coverage among the candidates provided by the newsmagazines (see Table 1).

_____________________________

TABLE 2 ABOUT HERE ____________________________

Time Period

To evaluate the impact of the presidential debates, newsmagazines' publication schedules were broken into three time periods to test if editors' visual selections might have been influenced by the debate performances of the candidates. The first time period (Time l) included issues from Sept. 5 to Sept. 26; Time 2: October 3 to October 24; and Time 3: October 31 to November 21. As Table 2 illustrates, a significant difference did occur. However, more examination of the findings uncovered that the significance in the crosstab was contributed by the differences in number of candidate visuals, rather than by the differences in the number of visuals published by the newsmagazines during the three stages of the campaign (X2 = 2.72, df=2, p>.05).

_____________________________

TABLE 3 ABOUT HERE

_____________________________

One interesting observation, however, was that only one picture of Bentsen appeared before the vice-presidential debate, while nine appeared the two weeks immediately following the vice-presidential debate. During the last time period, only three Bentsen visuals appeared, and those appeared in the campaign wrapup issue.

Size of Visuals

An evaluation of the mean picas of coverage received by the candidates showed statistically significant differences. Analysis of variance was used for three different tests. In the first test, the size of the picture was the dependent variable, while candidates and the three newsmagazines were the independent variables. Findings indicate that significance in the model was contributed only by the candidates, and no interactions were evident (See Table 3).

_____________________________

TABLE 4 ABOUT HERE

_____________________________

Using the Scheffe test to analyze the significant F score provided by the candidates, one significant difference was found in the size of the photos used for Bush and Dukakis. The Bush visuals were significantly larger (Mean difference = 237.3, S = 217.923, p< .05). Knowledge of publication was not a factor in the test.

In the second ANOVA test, photo size was the dependent variable and the independent variables were candidates and the three constructed time periods Results from the second test (see Table 4) were similar to the first as the candidate dimension contributed the only significant F in the model. Use of the Scheffe test again revealed that Bush visuals were significantly larger than those of Dukakis (Mean difference = 237.3, S = 215.67, p<.05). Knowledge of time periods was not a factor in the test.

_____________________________

TABLE 5 ABOUT HERE

__________________________

In the third ANOVA test, photo size was again the dependent variable and the independent variables were the three newsmagazines and the three time periods revolving around the two presidential debates. In this analysis, a significant interaction (see Illustration 1) was found between the magazines and the time periods. Scheffe tests were used to analyze the significance, and the findings showed that U.S. News & World Report published significantly larger pictures during the final time period after the second presidential debate than did the other two magazines. However, the larger mean size of photos in U.S. News was not the result of more pictures than the other two magazines during the last time period. On the contrary, U.S. News carried only 18 photos in issues involved in the last time period (Time carried 49 and Newsweek 25), but the size of those pictures was significantly larger than pictures published by the other two magazines.

Their coverage of the candidates during that period was virtually the same, in that both candidates received the same picture size of coverage. Over the three time periods, mean size of pictures in Time increased over three time periods, although not significantly; while picture size in Newsweek magazine fluctuated over the three time periods (see Table 5).

_____________________________

TABLE 6 ABOUT HERE

_____________________________

Candidates

On the individual variables, three significant differences were found in the 15 comparisons between the two presidential candidates (see Table 5). On facial expression, Bush was presented more often with a cheerful, confident look (t=3.87, df=225, p<.01). Also Bush was presented more often than Dukakis in half page or larger visuals (t=3.00, df=204, p<.01). The camera angle for Bush more often looked up to the candidate (t=2.16, df=227, p<.05).

The difference in size and camera angle was reflected in the significant difference for the perspective index (t=2.62, df=227, p<.01) which showed Bush visuals being treated more favorably than the Dukakis visuals. The other two indices--behavior and context--did not provide any significant differences between the two candidates.

Only one difference was statistically significant in the comparisons between the vice presidential candidates. Bentsen was more likely to be presented as the center of attention while Quayle was more frequently presented with a crowd or colleagues paying attention to something or someone else.

There was one other difference which was apparent in the content analysis but masked by the statistical tests (because of small cell size) and that occurred in the presentation of family association. Quayle was often depicted with family or spouse, while Bentsen seldom was. Mrs. Bentsen and

the Bentsen family were the missing persons in the 1988 presidential campaign. This is better depicted in Table 6.

In summary, the three newsmagazines displayed few differences in how they covered the campaign. Presidential candidates received significantly more coverage than did vice-presidential candidates in all three publications, and Republican candidates were pictured more often than Democrats. The number of pictures for each candidate differed significantly from each other, but the presidential debates did not have any effect on the number of pictures published by each publication. U.S. News & World Report campaign pictures were significantly larger in the last time period after the second debate, but the both presidential candidates received the same mean size play in that magazine during that time period.

Overall, George Bush's campaign photos were significantly larger than those of Michael Dukakis. Larger Bush photos appeared in all three publications, and the size of those photos was not influenced by the three stages of the presidential campaign. Bush received better page position; he appeared more cheerful in his visuals; and he received better camera angles than did Michael Dukakis. In the vice-presidential campaign, Lloyd Bentsen visuals were virtually absent from the coverage.

In general, the newsmagazines tried to provide balanced visual coverage of the presidential and vice-presidential candidates in the l988 campaign. In doing this content analysis, it was apparent that the editors were trying to match visuals as much as possible, even pairing them in many layouts, and this was particularly evident in U.S. News & World Report in the last time period.

It is true, however, that the Republican candidates, who were also the winners, did receive larger play. Bush appeared in more visuals and the visuals were larger than those used for Dukakis. Furthermore, Bush was depicted as more cheerful and confident than Dukakis and seemed to benefit from a more respectful camera angle. This Bush advantage carried throughout the 1988 campaign, which was much different from the way both his and Ronald Reagan's visual portrayal fluctuated in the l984 campaign.

Vice presidential coverage in l988 was in sharp contrast to coverage in l984. The interest that Geraldine Ferraro generated as the first woman to seek the vice presidential nod was mirrored in l984 coverage as she was portrayed more positively and in more photos than George Bush by the end of the campaign. In l988, newsmagazines did not seem interested in the vice-presidential campaign since they significantly underplayed both candidates. Quayle, who some seemed to think was being kept away from the cameras, did receive more coverage than Bentsen. The shots, however, were more likely to depict scenes where Quayle was an observer rather than the center of attention.

Bentsen went virtually unnoticed by the newsmagazines, and while probably underused by his party, he received very little coverage until the vice-presidential debate. This also was reflected in the strange pattern of coverage that eliminated Mrs. Bentsen and the Bentsen family from newsmagazine pages. All of the other candidates were depicted at various times with family and wives, except for Bentsen. Although the selection of Bentsen as a running mate was considered by many to have been a good Dukakis choice, it was obvious that the newsmagazines did not see it the same way.

While on the surface there seemed to be an effort made by the newsmagazines to provide balanced coverage, and there were few differences on most of the 15 dimensions evaluated, there is still evidence that one candidate, in this case Vice President Bush, received more and larger visuals as well as more favorable placement. Bush's advantage was subtle, and it was an advantage that was carried by all three newsmagazines. It was in place at the beginning of the campaign, and it did not waver throughout the three stages of the campaign. In this respect, there was evidence of a differential pattern of coverage apparent in the visuals used in newsmagazines in the 1988 campaign.

1. Bruce H. Westley, et. al., "The News Magazines and the 1960 Conventions," Journalism Quarterly 40:525-531, 647 (1963).

2. Guido H. Stempel III, "Prestige Press Meets the Third Party Challenge," Journalism Quarterly 46:701 (1969).

3. Dru Evarts and Guido H. Stempel III, "Coverage of the 1972 Campaign by TV, News Magazines and Major Newspapers," Journalism Quarterly 51:646-648 (1974).

4. "TV Study: Jackson Best, Robertson Worst," Boulder Daily Camera, March 15, 1988, p. 3.

5. Sandra E. Moriarty and Gina Garramone, "A Study of Newsmagazine Photographs of the 1984 Presidential Campaign," Journalism Quarterly 63:728-734 (1986).

6. Wayne Wanta, "The Effects of Dominant Photographs: An Agenda Setting Experiment," Journalism Quarterly 65: 107-111 (1988).

7. O. Holsti, Content Analysis for the Social Sciences and Humanities (Reading, MA: Addison-Westley, 1969).

8. Klaus Krippendorff, Content Analysis: An Introduction to its Methodology (Beverly Hills, CA: Sage Publications, 1980), p. 138.

9. Elizabeth Loftus and Geoffrey Loftus, Essence of Statistics (New York: Alfred A. Knopf, 1988), pp. 482-484.

10. Mean picture size for U.S. News & World Report during Time 3 was 1138.20 sq . picas; Newsweek, 397.84; Time, 588.90. Scheffe tests were signficant between U.S. News and Newsweek (MD=740.36, S=400.71, p<.05); and between U.S. News and Time (MD=549.30, S=353.38, p<.05).

EVALUATION ATTRIBUTES

more favorable less favorable

activity : dynamic behavior such as speaking, lethargic or passive activity such as

shaking hands, kissing babies listening, reading, dozing

posture: standing tall and upright bowed, slumped, or leaning on something arms: arms head high or above arms at side, at rest, folded

hands: gesturing or doing something hands at side, or at rest

eyes: eyes looking directly at camera or eyes up, down, or closed

at someone

expression: cheerful or confident unhappy, worried, or tired

interaction: cheering crowd or attentive colleagues candidate alone or with inattentive crowd or colleagues

camera angle: looking up at candidate looking down on candidate

portrayal: present the candidate as dignified, a presentation that made fun of the

serious, or presidential candidate, comic or goofy treatment

position: top of the page bottom of the page

size: larger than 1/2 page smaller than 1/2 page.

IMAGE ATTRIBUTES

props: coded as either campaign symbols like flags, bunting and the presidential seal;

or knowledge symbols like charts, graphs, or briefcases

setting: either monumental and formal; or informal and casual

dress: either dignified suit and tie; or sportclothes or shirtsleeves

family association: either with family, with spouse, or no family.

____________________________________________________________________

Time U.S. News Newsweek Total/Ave

n % n % n % n %

Presidential Candidates:

Bush 48 38 42 34 35 28 125 44

Dukakis 44 41 34 32 29 27 107 38

Vice-Presidential Candidates:

Quayle 11 28 16 41 12 31 39 14

Bentsen 5 39 4 31 4 31 13 5

Totals/Average 108 38 96 34 80 28 284 100

____________________________________________________________________

X2 = 2.202, df=6, p>.05

____________________________________________________________________

Time 1 Time 2 Time 3 TOTALS

N % N % N % N %

Presidential Candidates:

Bush 41 14 33 12 51 18 125 44

Dukakis 29 10 36 13 42 15 107 38

Vice-Presidential Candidates:

Quayle 21 7 10 4 8 3 39 14

Bentsen 1 .4 9 3 3 1 13 5

TOTALS 92 32 88 31 104 37 284 100*

____________________________________________________________________

X2 = 20.72, df=6, p<.01

*Because of rounding error, totals might not add up to 100

____________________________________________________________________

Sum of Mean Sig.

Source of Variation Squares DF Square F of F

Main Effects 6422511 5 1284502.11 3.659 .003**

Candidates 4957449 3 1652483.13 4.708 .003**

Publication 1466949 2 733474.31 2.090 .126

Interactions 1820995 6 303499.21 .865 .521

Candidate/Pub 1820995 6 303499.21 .865 .521

Explained 8243506 11 749409.62 2.135 .018

Residual 95474124 272 351007.81

TOTAL 103717630 283 366493.3

____________________________________________________________________

**p<.01

(seems to be missing a table--old #4, now #5: Mean Picture Size by Candidate and Time Period)

____________________________________________________________________

Sum of Mean Sig.

Source of Variation Squares DF Square F of F

Main Effects 4607677 4 1151919.372 3.357 .011

Publication 2144264 2 1072182.133 3.124 .046

Time Periods 3142616 2 1571308.167 4.579 .011

Interactions 4734070 4 1183517.377 3.449 .009

Pubs/Time 4734070 4 1183517.377 3.449 .009

Explained   9341747 8 1167718.374 3.403 .001

Residual 94375883 275 343185.029

TOTAL 103717630 283 366493.392

____________________________________________________________________

____________________________________________________________________

Bush Dukakis Quayle Bentsen

n=125 n=107 n=39 n=13

BEHAVIOR

Activity 2.53 2.65 2.54 2.54

Posture 2.70 2.60 2.59 2.77

Arms 2.14 2.12 1.97 2.00

Hands 2.52 2.56 2.33 2.31

Eyes 2.30 2.36 2.41 2.38

Expression 2.41 2.15** 2.44 2.46

TOTAL 14.48 14.43 14.28 14.46

CONTEXT

Props 1.99 1.96 2.05 2.23

Setting 1.78 1.84 1.90 1.85

Dress 2.62 2.44 2.51 2.69

Interaction 2.01 2.08 1.95 2.46**

Family 1.98 1.94 2.03 2.00

TOTAL 10.38 10.29 10.44 11.23

PERSPECTIVE

Position 2.18 2.06 2.46 2.38

Size 1.33 1.11** 1.03 1.00

Camera Angle 2.18 2.03* 2.13 1.92

Portrayal 2.02 1.99 1.97 2.23

TOTAL 7.71 7.20** 7.59 7.54

____________________________________________________________________

**p < .01

*p < .05

_____________________________________________________________

With Spouse Alone/With Colleagues With Family

Bush 8 112 5

Dukakis 7 99 1

Quayle 4 30 5

Bentsen 1 12 0

_____________________________________________________________

______________________________________________________

Sum of Mean Sig.

Source of Variation Squares DF Square F of F

Main Effects 6879060 5 1375812.006 3.971 .002

Candidates 4415747 3 1471915.603 4.249 .006

Time Periods 1923498 2 961749.054 2.776 .064

Interactions 2609125 6 434854.185 1.255 .278

Cand/Time 2609125 6 434854.185 1.255 .278

Explained 9488185 11 862562.285 2.490 .005

Residual 94229445 272 346431.782

TOTAL 103717630 283 366493.392

____________________________________________________________________

clickhere to return to Professor Moriarty's home page

click here to return to the visual communication research page