This paper was presented at the Association for Education in Journalism and
Mass Communication in San Antonio, Texas August 2005.
If you have questions about this paper, please contact the author
directly. If you have questions about the archives, email
rakyat [ at ] eparker.org. For an explanation of the subject line,
send email to
[log in to unmask] with just the four words, "get help info aejmc," in the
body (drop the "").
(Feb 2006)
Thank you.
Elliott Parker
====================================================================
Comparing Two Kinds of News Reports about Political Ads:
A Model to Predict Candidate Evaluation
Lisa Mills
University of Florida
(407) 823-5671
[log in to unmask]
College of Journalism and Communications
2045 Weimer Hall
University of Florida
Gainesville, FL 32611
Pilar Bermudez
[log in to unmask]
University of Central Florida
(407) 823-2000
Nicholson School of Communication
P.O. Box 161344
Orlando, FL 32816
A/V REQUIREMENT: Overhead Projector
Abstract
Researchers conducted experiments to compare television news viewers'
perception of political ads, traditional political race profile
reports and ad watch reports. Utilizing real ads and television news
reports surrounding competing candidates in a 2002 Congressional
Race, researchers found evidence to support their hypothesis that
viewers perceive ad watches differently from traditional race profile
reports that do not attempt to examine ad truthfulness. Subjects
found ad watch reports clearer and easier to understand. Researchers
also found evidence to support their hypothesis that viewers'
feelings about candidates differ, depending on whether they are seen
in political ads, race profiles or ad watch reports. A model to
predict candidate evaluation brought mixed results.
Comparing Two Kinds of News Reports about Political Ads:
A Model to Predict Candidate Evaluation
People are watching political ads in commercial breaks and on the
news, but are they paying attention? Among those who do pay
attention, which is more likely to affect voting behavior, the ads or
news coverage of the ads? These are two questions to which both
political science and communication scholars seek answers. Every
election year brings an increase in the number of ads produced and
the amount of money spent making them and putting them on the
air[1]. Yet, news organizations spend less time each year reporting
on whether or not those ads are truthful. Collaborative studies
conducted by the NewsLab at the University of Wisconsin-Madison
during the fall 2002 and 2004 election seasons[2] found viewers
watching the most popular local newscasts in the largest metropolitan
markets saw twice as much political advertising as they saw
non-partisan news stories about the elections. How might this
affect voting behavior? Since the advent of electronic media,
numerous studies have explored the effects of political ads and news
political coverage on democratic communication (Berelson, Lazarsfeld
& McPhee, 1954).
Review of Literature
Media Effects and Voting Behavior
The origin of media effects research lies in the study of political
communication. Berelson et al's study of the 1948 election in Elmira,
New York, empirically supported many of the basic assumptions made
about voter behavior by both political and communication
scientists. It was the first to analyze campaign mass media under
Ernest Barker's analogy of a "system of discussion" (p. 41, inside
Berelson et al, 1954, p. 235) in which candidates supply the message,
the media distribute it, and the electorate "consume it" (Berelson et
al, 1954,p. 235). Many variables contribute either directly or
indirectly to the discussion, from political party rhetoric, to
communication channel selections and preferences to the personal and
social characteristics of the voters themselves. The Elmira study
established that the effects of these variables and many others must
be considered when conducting any study of political communication.
The advent of the electronic media brought great potential to expand
the "system of discussion" but voter information levels have remained
relatively low. Campbell, Converse, Miller and Stokes (1966) found
that even though political information is more accessible and voters
are often exposed to it, they do not pay much attention. They
conclude as more people are exposed to political messages in the mass
media, they are less likely to discuss those messages on the
interpersonal level. This creates a vacuum in which, despite the
strong flow of information from the mass media, no opinion forms at all.
More recent research on public opinion establishes there is a strong
relationship between people's level of political awareness and their
reception to news stories. Zaller (1992) finds the more familiar the
issue in a news story, the less overall attitude change. However,
those news viewers who do change their attitudes are more likely to
be less aware. The level of intensity of the message is also a
factor because it can produce a higher or lower proportion of
change. Zaller cites Shipler's 1986 New York Times article about a
poll that found attitude change on whether the United States should
back the Contra guerillas in Nicaragua depended on a message of
middle intensity and lower levels of issue familiarity.
Since the mass media have become such an integral part of the
political campaign process, the question, "do campaigns matter?"
often sends researchers back to media effects studies. Communication
and political scientists want to know how much voters learn from such
campaigns, and whether those campaigns can affect voter
behavior. Much has been learned and still more debated about this
process. Popkin (1994) believes voters can make rational decisions
based on very little political information. Bartels (1996) contends
uninformed voters find "shortcuts" through the forest of political
information. Page and Shapiro (1992) might argue that even if
individuals make random votes that change from issue to issue
independent of party identification, aggregate level opinions remain
stable over time.
How then, does the combination of political campaign advertising and
news media coverage of such affect voter behavior? These two aspects
of political communication come together in the study of a specific
type of news report called an "ad watch."
Political Ads and Ad watches
Presidential candidates in 1956 were the first to spend millions of
dollars on televised political ads but it was not until 1964 that a
political ad made national news. Lyndon Johnson's Daisy Girl aired
as a paid spot only once but was replayed in its entirety on all
three network newscasts because of its controversial images and fear
appeal (Jamieson, 1992; West 1993; Kaid, McKinney & Tedesco,
2000). As television became the dominant mass medium and candidates
increasingly used it to communicate directly to voters, news
organizations increased their coverage of political ads. News
organizations analyzed political ad content in their stories with
varied methods and styles, and these reports generally became known
as "ad watches." A content analysis of televised presidential ad
watches appearing on network newscasts revealed that between 1980 and
1988 the number of ad watches tripled, and there were six times as
many ad watches in 1988 as in any other presidential election year
(Kaid, Gobetz, Garner, Leland & Scott, 1993). Kaid's research and
that of other political communication scholars in the early 1990s
also revealed four major reasons why television news organizations
are likely to report on the content of political ads. First, they
provide an easy source of visuals and conflict, both essential
elements for compelling news stories (Kaid et al., 1993; Jamieson,
1992). Second, they provide a direct channel of communication
between candidate and voter that could be mediated through news
reports (Kaid and Johnston, 1991). Third, ad watches fit into the
television news format of horse race political reporting (Kaid et
al., 1993). Finally, a side effect of ads is that their content set
the agenda for which campaign issues will receive media coverage
(Kaid et al., 1993). Ad watches' agenda-setting function will be
discussed later in this paper.
A network news ad watch of a Social Security ad by Dukakis and a
tank ad by Bush marked a turning point in 1988. Jamieson (1992)
conducted an experiment revealing that viewers remembered the ads
rather than the corrections offered by ABC reporter Richard
Threlkeld. The experiment offered evidence of something Jamieson and
other scholars had suspected, that ad watches could backfire and
enhance the message of the ad. So Jamieson, a research team from the
University of Pennsylvania's Annenberg School, and CNN reporter
Brooks Jackson set out to build a better ad watch, one designed to
prevent this backfire effect. They developed a visual grammar for ad
watches, utilizing three specific formatting techniques designed to
encourage the viewer to process the content of the ad watch, rather
than the content of the ad: distancing placed the ad inside a mock
television screen; disclaiming attached a news logo and a notice that
the visuals are part of an ad for a particular candidate; displacing
meant the reporter interrupted the ad to comment on its content and
to place graphic "correctors" on the screen (Capella & Jamieson,
1994). Subsequent experimental research utilizing these formatting
guidelines brought mixed results. Cappella and Jamieson (1997)
believe that although the visual grammar for ad watches is not a
cure-all, it works. Others continue to argue any ad watches have the
capacity to backfire (Pfau & Louden, 1994; Ansolabehere & Iyengar,
1996; Just, Crigler, Alger, Cook, Kern & West, 1996). Television
networks seemed to back away from ad watches between 1988 and 1992,
when Kaid, Tedesco and McKinnon (1996) found political ad coverage
dropped by 50 percent. Kaid, McKinney and Tedesco (2000) report a
slight increase between 1992 and 1996, but the more recent Lear
Center study indicates a new decline. Mills (2003) found that during
the 2002 Florida Gubernatorial election, only two of all the
television stations in medium or large Florida markets produced ad
watches on any race.
News Coverage of Political Ads
Journalists have special access to political information. They can
use their credibility as investigators to evaluate claims in
political ads and serve the public by reporting on political ads that
make deceptive or misleading claims. Politicians may then create and
air ads to respond to such reports and thus "reframe" the information
(Jamieson, 1992).
Inside the broadcast paradox broadcasters, using a public resource
in the form of electromagnetic waves, try to balance commercial
interests with public service obligations. On the local level, that
boils down to station managers (under increasing corporate pressure
for profit) having the difficult task of balancing political
advertising policy with community needs. Corporate ownership is also
a factor. Results of the 2002 Lear Center study indicated stations
with large owners provided less political coverage than stations with
small and mid-sized owners. Ad watches utilizing Jamieson's visual
grammar guidelines require extensive research by reporters and
time-consuming technical production. A 2002 survey of 103 news
directors conducted by the Project for Excellence in Journalism[3]
found the average local news reporter is expected to produce two
different news stories each day, the highest ever in five years of
research. This may provide one explanation for the decline in
political ad reporting found by the Lear Center. Interestingly, a
2003 survey of local television news directors by the Journalism
Ethics Project[4] finds 27 percent believe political news reporting
on their station is "excellent" and 56.6 percent believe it is
"good." When asked, "do you think those reports help people make up
their minds about which candidate they prefer?" 82.4 percent said "yes."
While scholars hope that the analysis of political ads on television
news can be helpful to voters, as Kaid notes, "so far the media have
done little to prove themselves worthy of the journalistic
responsibility to scrutinize what may be the most important aspect of
campaign discourse" (Kaid, 2004, p. 188). Jamieson warns that "the
sheer volume of political advertisements can easily overshadow the
balanced news coverage a station provides" (Jamieson, 1999, p.
15). Political ads provide candidates the ultimate tool in
controlling their messages to voters. Television stations receive
revenue in exchange for providing a tremendous reach and penetration
of these messages. Should they not, then, distribute an objective
counterbalance through their news departments? If this public
service obligation were not enough, the following summary of
political ad and ad watch effects may provide another convincing
argument for news coverage of political ads.
Theoretical Perspective
A complete discussion of direct and limited effects theory would be
too lengthy for this paper. However, it would be useful to briefly
outline some of the major theoretical underpinnings for the study of
political ads in order to establish the need to for further studies
about media coverage of them. Research under the limited effects
paradigm indicates that political ads affect voters cognitively,
affectively, and behaviorally.
As early as 1976, Patterson and McClure found evidence that voters
learned about issues from political ads. In fact, their study of the
1972 presidential election found political ads had more content about
issues than network newscasts. Twenty-five years later, an analysis
of the 1996 presidential election confirmed that political ads have
cognitive affects (Lichter & Noyes, 1996). There is plenty of
evidence that voters not only learn more from political ads than from
television news, but also more than from televised debates (Just,
Crigler & Wallach, 1990). The format of ads can affect voter recall
(Kaid & Sanders, 1978) and that voters generally remember the content
from negative ads better than from positive ones (Basil, Schooler &
Reeves, 1991).
A political ad is said to have affective effects if it changes
the way voters feel about candidates (Kaid, 1981). While the
evidence of this has been less convincing than that for cognitive
effects, some studies conducted on lower elections levels indicate
televised political ads can be correlated with candidate evaluations
(Atkin & Heald, 1976; Mulder, 1979). Numerous studies indicates
political ads can change the way a voter feels about a candidate
(Atkin & Heald, 1976; Hofstetter, Zukin & Buss, 1978; Kaid &
Chanslor, 1995; Kaid, Leland & Whitney, 1992; West, 1993). Political
ads that focus on issues seem to succeed in changing the image of the
candidate in the minds of the voters (Kaid, Chanslor & Hovind,
1992). Agenda-setting effects have been associated with political
ads, primarily in helping to explain their cognitive
effects. Political ads featuring issues have been shown to affect
issue salience for voters and the news agenda of media outlets (West,
1993; Schleuder, McCombs & Wanta, 1991).
Behavioral effects of political advertising could consist of anything
from voting intent to contributing to a political
campaign. Obviously, many variables must be considered and a causal
effect would be very difficult to prove. Hofstetter and Buss (1980)
were able to find a positive association between political ads that
air late in a campaign with voter turnout and changes in candidate
preferences. Meadow and Sigelman (1982) conducted an experiment in
which they exposed subjects to political spots featuring an actual
congressional candidate. They concluded, "the extent to which voters
can be manipulated by manufactured images may indeed be severely
restricted" (Meadow and Sigelman, 1982,p. 173-74). Ansolabehere and
Iyengar (1994) also chose an experimental design to test the effects
of "issue" political spots and their interaction with news
coverage. They were able to demonstrate issue framing in news is
important and could determine the success of "issue" political spots
(p. 355). Ansolabehere, Iyengar and Valentino made important
contributions to research on the "demobilization hypothesis," which
contends attack ads contribute to low voter turnout. They
constructed an experiment exposing subjects to both negative and
positive political spots and found, "exposure to campaign attacks
makes voters disenchanted with the business of politics as usual"
(1994, p. 835). A replication study using aggregate and survey data
produced the same result (Ansolabehere, Iyengar and Simon,
1999). However, Wattenberg and Brians believe the aforementioned
studies exaggerated the demobilization dangers posed by attack
advertising. They claim their evidence shows that "if negative
commercials persuade voters that the choice between the candidates is
an important one, then they are likely to increase rather than
decrease turnout" (1999, p. 896). Goldstein and Freedman also found
evidence that exposure to negative campaign ads can actually
stimulate voter turnout (2002). The debate continues over the
effects of negative political spots, and most research has centered
on ads that were negative, or in which one candidate attacked their
opponent. However there is at least one consistent finding. Most
research on cognitive, affective and behavioral effects seems to
indicate political advertising is more effective when the level of
voter involvement is low (Kaid, 1981) or mid level (Zaller, 2002).
While political ads set the agenda of television news, television
news legitimizes political ads through its judgment of their
significance (Roberts & McCombs, 1994). Researchers want to know
whether ad watches can counterbalance political ads with the same
cognitive, affective, behavioral and agenda-setting
effects. Experimental studies continue to bring mixed results.
If remembering the content of an ad watch can be considered learning
from it, then ad watches do have a cognitive effect, since viewers do
recall what they see and hear in ad watches (Cappella & Jamieson,
1994). However, this cognitive effect may be tied to the debate over
the affective effect of ad watches. This debate continues. Some
scholars say there is evidence that ad watches work because they help
voters decide which political ads are fair and which issues featured
in those ads are important (Cappella & Jamieson, 1994). Others
present findings that indicate ad watches do not work, because they
instead enhance the message of the campaign ad being analyzed, and
thus the voter's opinion of the sponsoring candidate (Pfau & Louden,
1994; Ansolabehere & Iyengar, 1996; Just et al, 1996; McKinnon &
Kaid, 1999). Cappella & Jamieson (1997) believe that it is difficult
to determine the exact affective effects of ad watches because they
are so complex and difficult for voters to process, and they continue
to advocate ad watch formats that simplify political ad critiques.
Kaid's content analysis of ad watches indicate they perform an
agenda-setting function, due to their placement generally within the
first ten minutes of network newscasts (Kaid, Gobetz, Garner, Leland
and Scott, 1993). Under inoculation theory, scholars argue that if
newscasts aired carefully produced ad watches prior to voters'
exposure to negative political ads, voters might be better able to
resist persuasive appeals in the ad (Cappella & Jamieson, 1994;
O'Sullivan and Geiger, 1995). This researcher was unable to locate
any research that has been conducted to determine whether ad watches
have any effects on actual voter behavior.
Purpose of this study
Established research on political ads and ad watches calls for
future experimental studies to test differences between cognitive and
affective effects of ads and ad watches. It calls for the continual
examination of the language and production techniques employed in ads
and ad watches and how both affect journalists' and viewers'
responses to the reports, the ads and the candidates. Media
professionals want to know whether they should devote the time and
space needed to thoroughly analyze political advertisements' claims
and to discuss any ethically suspect techniques.
This study will attempt to build on the research of prior scholars
who have used experiments to study the effects of political ads and
ad watches. A new contribution may be made through the comparison of
these effects with those of a more traditional news political report,
often called a "race profile." Such a report provides information on
candidates' background and their position on issues. But while the
report may use parts of political ads as background material or
"b-roll," it does not attempt to determine whether those ads are
accurate or misleading. The researchers' collection of the original
political ads, an ad watch featuring those ads and a traditional race
profile containing those same ads presents a unique opportunity to
study and compare their effects, thus the following research
hypotheses are made:
H1: Television news viewers perceive ad watch reports differently
than traditional political race profile reports.
H2: Ad watch reports will be easier to understand than traditional
race profile reports.
H3: Television news viewers' feelings about candidates will differ,
depending upon the format in which they are seen (political ad, ad
watch or race profile)
H4: A model can be used to predict opinion of candidate, based on
the format in which the candidate is seen (political ad, ad watch or
race profile) along with the viewer's age, gender and party identification.
The practical question looming here is how would the public best be
served? Should television journalists make the investment in
producing ad watches that utilize Jamieson's visual grammar? Or,
would they be just as successful if they produced the traditional
political race profile?
Method
Quasi-Experimental Design
The goal of this study is to contribute to the body of knowledge
about ad watch reports. One way to determine whether ad watches work
is to conduct experiments. Experiments can measure the effects of ad
watches and other kinds of news reports about ads. The effects of
the ads themselves can also be measured. Comparative statistical
tests may reveal whether the traditional race profile, the ad watch,
or the ads produced the strongest effects on test subjects.
This study employed a multivariate posttest-only design with six
group exposures. Two groups were exposed to one of three instruments
(ad watch, race profile report or political ads) and a posttest
produced dependent variables (effects) in the form of ordinal data.
Test Groups
Subjects were chosen from undergraduate students enrolled in
speech classes at a large southeastern university. An opening
statement informed the students their grades would not be affected by
their participation in the experiment, and to insure confidentiality,
students were assigned a research number and asked not to identify
themselves on the questionnaire. While the external validity of
research conducted on students has often been questioned, there were
several reasons for using students for this study. First, they were
unfamiliar with the candidates featured in the news reports and
political ads. Second, studies have shown students may not have yet
formed clear political preferences (McKinnon, 1995). Finally, a
sample of students allowed the researcher to collect information
about political efficacy and media use among an age group that many
predict will have an important impact on upcoming national elections.
Experimental Stimuli
The political ads (Instrument C), the ad watch (Instrument B) and
the political race profile report (Instrument A) used in this study
actually aired in Orlando, Florida, during October and November, of
2002. They featured two candidates in the race for Congressional
District 24, a new district created by the Florida legislature,
meaning neither candidate was the incumbent. Republican and former
state house speaker Tom Feeney was the favorite, with a number of
years experience as a state lawmaker and plenty of GOP money in his
campaign coffers. Democrat and personal injury attorney Harry Jacobs
was a long shot candidate who gained media exposure during the
notorious 2000 presidential campaign controversy in Florida. Jacobs
filed an unsuccessful lawsuit against the Seminole County Elections
Supervisor over the validity of hundreds of absentee ballots. Each
candidate produced both positive and negative political spots during
the course of the campaign. These spots often aired inside
commercial breaks of local newscasts. Two of these breaks were
collected by the researcher, one containing a negative ad targeting
Feeney, and one containing a negative ad targeting Jacobs. These
breaks were edited into a fifteen-minute newscast containing real
content from the Orlando NBC affiliate.
The researcher collected the ad watch and the traditional political
race profile report containing the same two ads by Feeney and Jacobs
from two separate newscasts on the Orlando NBC affiliate. The same
reporter, Greg Fox, did both reports. Each was edited into the
identical fifteen-minute newscast employed for the political spots,
except this time the commercial breaks did not contain either of the
spots. The researcher feared placement of the reports could have an
affect on the dependent variable, so neither report was placed at the
end of a newscast segment, going into the commercial break. Instead,
both reports were consistently placed in the middle of the "A block"
of the newscast.
Careful examination of the Feeney and Jacobs spots reveals they
contain rhetoric and production techniques that would probably be
considered "classic" among political ad experts. The ad watch and
political news report produced by Greg Fox both contain production
techniques recommended by the Radio Television News Directors
Association. These techniques employ Kathleen Hall Jamieson's visual
grammar for ad watches, as described earlier.
Questionnaire
The questionnaire given to subjects after they were exposed to
selected experimental stimuli contained four basic parts and utilized
four different kinds of measurement.
The first part contained a series of semantic differential scales
adapted from a study that compared broadcast with print ad watches
(McKinnon, 1995). These scales were designed to collect responses on
the reports about the race profile report (Instrument A), the ad
watch (Instrument B) and the ads themselves (Instrument C). A second
set of scales was designed to collect responses to the candidates
themselves, in terms of credibility and image. Subjects also
responded to a "feeling thermometer" for each candidate.
The second part of the questionnaire contained multiple-choice
questions in order to determine how much respondents learned about
the candidates from the reports and ads. Respondents were also asked
whether or not they felt the candidates were being truthful and
whether they were likely to vote for either of the men.
The third part of the questionnaire contained traditional five-point
Likert scales to assess feelings about politicians, political ads,
television news coverage of ads and general political efficacy.
The last part of the questionnaire contained forced-choice questions
designed to collect demographic and media use information, along with
party identification. The final questions also asked subjects to
assess the importance of political advertising and media analysis in
their decisions about voting.
Procedure
At the beginning of each class students were read
instructions. Lights in the classroom were dimmed and a videotape
shown (Instrument A, B or C). After the video, the lights were
turned on and students were handed a questionnaire to complete. Most
students were able to finish the questionnaire in the time allotted
by the class instructor, but not all. The experiment was limited to
35 minutes. The questionnaires were collected and students were told
the nature of the experiment. Data from the questionnaires were
entered into SPSS datasheets by the researchers and analyzed as
described in the next section.
Findings
Test Groups
As discussed earlier, six speech classes were divided into three
test groups depending upon the experimental stimulus. Those groups
were fairly evenly distributed in terms of gender, median age and
party identification, with one exception. The group that saw the
race profile had distinctively more Republicans (40%) than Democrats
(20%). The fact that in each group and the test population 40
percent or more reported being Independents or "don't know" would be
expected for this age group, and could be considered advantageous for
a study of this nature. Many college students have not yet formed
clear political preferences, which, as McKinnon (1995) points out
"may make them less susceptible to partisan politics and better able
to provide candidates impressions" (p. 53).
Perception of Race Profile and Ad Watch Reports
Before examining the effects of the race profile, ad watch or ads
upon candidate evaluation, the researchers sought support for
hypotheses 1 and 2. A seven-point semantic differential scale using
bi-polar adjectives asked subjects to rank reports on whether they
were perceived as clear, informative, fair, understandable,
revealing, balanced, truthful, interesting, beneficial, good,
exciting, accurate or believable. While there was minimal evidence
that overall, subjects perceived the ad watch report differently than
traditional political race profile report, stronger evidence suggests
subjects found the ad watch clearer and easier to understand. Table 1
indicates subjects ranked the ad watch significantly higher on
clarity and understandability thus supporting hypothesis 2:
Table 1
Comparisons of Report Format (Profile, Ad Watch) on Evaluations of Report
Race Profile
(n =46)
Ad Watch
(n=44)
Clear?*
M=4.52
sd=1.09
M=5.71
sd=1.45
Understandable?*
M=4.63
sd=1.35
M=5.52
sd=1.45
*t tests significant at p<.05
Opinion of Candidate
Two new variables were created as composite measures of subjects'
evaluations of each candidate. For both Jacobs and Feeney, subjects
responded to the same seven-point semantic differential scale using
bi-polar adjectives. Responses were averaged to create the "Feeney
Evaluation Index" and "Jacobs Evaluation Index." Multivariate
analysis determined whether subjects' feelings about the candidates
were associated with the format in which they were seen. Results are
listed in Table 2.
Table 2
Opinion of Candidates
Format
Feeney Index
Jacobs Index
Race Profile
(n=46)
M=4.34
sd=.83
M=4.13
sd=.78
Ad Watch
(n=44)
M=4.07
sd=.76
M=4.15
sd=.85
Ads Only
(n=45)
M=3.70
sd=.75
M=3.72
sd=.78
Gamma
-.381*
-.278*
*Gamma is significant at the p<.01 level.
There is a significant relationship between subjects' opinion of each
candidate and the format in which they were seen, thus supporting the
third hypothesis. Gamma tests indicate a moderate negative
association between opinion of candidate and format for both
candidates. Knowing the format in which Feeney was seen would
improve our estimate of subjects' opinion of Feeney by 38
percent. Knowing the format in which Jacobs was seen would improve
our estimate of subjects' opinion of Jacobs by 27 percent. Knowing
the values placed in each format category (1=Profile, 2= Ad Watch, 3=
Ads Only), the negative association here tells us that the more
likely the candidate is seen in ads alone, the lower the opinion
subjects will have of him.
Predicting Opinions of Candidates
The researchers hypothesized that a model could be created through
multiple regression in order to determine which demographic and
composite measures might best predict subjects' overall opinion of
the candidates. Prior research supports viewer opinions based on
credibility and image. Thus, two more variables were created as
composite measures for the model.
The first new variable henceforth referred to as the "Truthfulness
Index," averaged selected responses in the semantic differential
scales where subjects evaluated the two different reports containing
the ads. When scales measuring "truthfulness" and "accuracy" were
averaged, a Cronbach's Alpha value of .6744 indicated they would be a
valid composite measure. Likewise, combining those same two scales
where subjects evaluated the political ads alone produced a
Cronbach's Alpha value of .7871.
The second new variable henceforth referred to as the "Watchability
Index," averaged scales measuring "interest" and
"excitement." Cronbach's Alpha value of .8389 indicated this
composite measure would be valid as a composite measure for the news
reports. Cronbach's Alpha value of .7862 indicated it would also be
valid as a composite measure for the political ads.
A theoretical basis exists for the creation of a prediction model
that would include explanatory variables such as age, gender and
political party affiliation. This study has found some evidence that
the format in which a candidate is seen (race profile, ad watch or
political ads alone) has a moderate association with subjects'
opinion of a candidate. As mentioned earlier, prior research
suggests candidates are evaluated on the basis of viewers'
perceptions of credibility and image. Thus, the researcher
formulated the prediction equation "opinion of candidate" =
constant + b1(Age) + b2(Gender) + b3(political party ID) +
b4(format test group) + b5(report truthfulness index) + b6(report
watchability index) + b7(ad truthfulness index) + b8(ad watchability
index) + error.
Figure 1 contains a scatterplot of conditional distributions about
the mean opinion of Feeney, revealing a linear association among the
8 predictors.
Figure 1
Linear Regression: Opinion of Feeney
In performing linear regression for this model, the null hypothesis
would be H0: b1+b2+b3+b4+b5+b6+b7+b8=0. The alternative hypothesis
would be that at least one of the predictors has an effect on the
opinion of Feeney. The overall model's F-value of 5.428 yields an
extremely small P-value, so the null hypothesis is rejected. At
least one of the predictors has an effect on opinion of Feeney. The
multiple correlation coefficient (R=.616) indicates that when using
the selected explanatory variables to predict opinion of Feeney, for
every one standard deviation increase in the prediction equation
there is a .616 increase in opinion of Feeney. The multiple
coefficient of determination (R Squared = .379) indicates that when
using the prediction equation to predict opinion of Feeney instead of
the mean opinion of Feeney, the proportional reduction in error is 38
percent. Table 3 depicts the results of the first of two prediction
equations, one for each candidate.
Table 3
Linear Regression of Opinion of Feeney on 8 Predictors
Predictor
Slope
Standard Error
Standardized Coefficient (Beta)
t value
P
Age
.009184
.050
.190
1.837
.070
Gender
.249
.155
.154
1.610
.112
Party ID
-.00189
.068
-.028
-.279
.781
Format
-.201
.151
-.129
-.1333
.187
Report Truthfulness
.005977
.046
.138
1.306
.196
Report Watchability
.0001122
.041
.003
.028
.978
Ad Truthfulness
.264
.069
.415
3.835
.000
Ad Watchability
.006469
.066
.122
.986
.328
An examination of the standardized coefficients in this model
indicates that Ad Truthfulness is the strongest and only
statistically significant predictor (Beta =.415). Age (Beta =.190),
Gender (Beta = .154), Report Truthfulness (Beta =.138) and Ad
Watchability (Beta = .122) follow in that order, with weak positive
associations. This indicates that women are slightly more likely to
think highly of Feeney, and opinion of this candidate increases as
age, report truthfulness and ad watchability also increase. Format
has a weak negative association, meaning that the more likely
subjects were exposed to Feeney in political ads, the less subjects
are likely to think of him. Party identification and report
watchability have almost no prediction value in this model. When the
researcher removed them from the model the Betas were barely
affected, indicating neither party identification nor report
watchability had any strong interactions with other variables, nor
did they suppress the effect of other variables.
Figure 2 contains a scatterplot of conditional distributions about
the mean opinion of Jacobs, revealing a slightly smaller linear
association among the 8 predictors.
Figure 2
Linear Regression: Opinion of Jacobs
Again, in performing linear regression for this model, the null
hypothesis would be H0: b1+b2+b3+b4+b5+b6+b7+b8=0. The alternative
hypothesis would be that at least one of the predictors has an effect
on the opinion of Jacobs. For this candidate, the overall model's
F-value of 1.481 yields a P-value that is not statistically
significant (.179) so here we cannot reject the null hypothesis. It
appears that the model used to predict opinion of Feeney is not as
good at predicting opinion of Jacobs. The multiple correlation
coefficient (R=.378) indicates that when using the selected
explanatory variables to predict opinion of Jacobs, for every one
standard deviation increase in the prediction equation there is a
.378 increase in opinion of Jacobs. The multiple coefficient of
determination (R Squared = .143) indicates that when using the
prediction equation to predict opinion of Feeney instead of the mean
opinion of Jacobs, the proportional reduction in error is only 14
percent. Table 7 depicts the results of the prediction equation for Jacobs.
Table 4
Linear Regression of Opinion of Jacobs on 8 Predictors
Predictor
Slope
Standard Error
Standardized Coefficient (Beta)
t value
P
Age
-.00662
.060
-.131
-1.094
.278
Gender
.008894
.190
.053
.467
.642
Party ID
.00117
.082
-.022
-.190
Format
-.00713
.184
-.044
-.387
.700
Report Truthfulness
.002108
.053
.047
.397
.693
Report Watchability
.00117
.046
.034
.253
.801
Ad Truthfulness
.203
.082
.308
2.473
.016
Ad Watchability
-.00312
.082
-.054
-.382
.704
Again, the index of ad truthfulness appears to be the strongest
predictor, but other results drawn from this model are inconclusive.
Discussion
Volumes of studies have been conducted trying to pinpoint causes of
candidate likeability and electability but that is not the central
issue of this research. Instead, the authors sought support for
hypotheses about television news political coverage, aiming to
provide insight into how viewers perceive different kinds of reports
about political ads compared with their perception of the ads
alone. A well-distributed sample provided subjects of college age,
generally not yet committed to one political party or
another. Indeed, political party affiliation did not significantly
affect any of the findings.
Race Profile vs. Ad Watch
Ad watches are complicated. They sometimes take weeks to research
and produce. The reporter must analyze and compare two thirty-second
ads in a report that often must run no longer than two or three
minutes. If the candidates have served as elected officials in the
past, as they often have, the reporter must conduct extensive
research to see if claims about their voting records and other
matters are indeed true. The reporter often gives each candidate the
opportunity for a rebuttal sound bite. In the case of the stimuli
utilized in this experiment, the reporter was extremely careful in
following Jamieson's visual grammar for ad watches in his production
techniques. These techniques used to be time-consuming, but new
digital editing systems are making them easier to achieve. Still,
most political reporters and producers would probably say an ad watch
could never be created in a day.
In contrast, political race profiles are produced routinely. The
reporter typically explains the office for which the candidates are
running, gives some background on those candidates, and includes a
sound bite or two from each candidate and some video or sound from
their political ads. A race profile report could certainly be put
together in one day.
This study found that the extra work and resources involved in
producing ad watches are worth the trouble because subjects found
them clearer and easier to understand. This study provides more
evidence that Jamieson's visual grammar for ad watches is a valid
production technique that assures the viewer understands which ad
claims are true and which claims are false. With the increasing
amount of political ads being shown during commercial breaks in
newscasts, ad watches could certainly serve the public interest in
helping viewers know which ads to believe.
Evaluation of Candidates
The researchers hypothesized television news viewers' evaluations of
candidates would differ, depending on whether those candidates were
seen in political ads, ad watch reports or traditional race profile
reports. This study supports that hypothesis, revealing a moderate
relationship between the format in which a candidate is seen and the
overall opinion of that candidate. When a candidate was seen in a
news report, either the profile or the ad watch, subjects' opinion of
him was likely to be higher. This finding provides more evidence
supporting the status conferral effect of television news. Campaign
managers would do well to have their candidates agree to talk to
reporters and appear on television in reports about their political
ads. These findings indicate viewers' opinions of the candidate
would likely be higher than if they saw the candidates in the ads alone.
Predicting Opinion of Candidate with Regression Models
The fourth and final hypothesis made by the researchers is supported
with one candidate, but not with the other. As with so much
political communication research, results are mixed. When the model
combines all variables with ad truthfulness it does a fairly good job
in predicting opinion of Feeney. While that model is not
statistically significant for predicting opinion of Jacobs, a linear
association does exist. Further revisions might reveal a better
model for predicting opinions of both candidates. However each
political campaign and each political candidate are unique. Therein
lays the biggest challenge in trying to predict who will win an
election and whether ads or coverage on television news will help or
hinder them.
The research model tells us something else that is worth noting. A
viewer's opinion of a candidate is probably most often based on
whether they think a candidate is telling the truth in their
political ads. The model finds statistical support for this
assumption, and it makes sense in today's mass media political
campaigns. Viewers are many times more likely to see a candidate in
their political ad than in person or even in a news report. Thus,
viewers are likely to transfer the ad's level of honesty directly to
the candidate.
Finally, this study provides evidence that while producing ad
watches would indeed serve the public interest, those news reports
about the ads probably won't do as much to change viewers' opinions
as the ads themselves. Reports inside one or two thirty-minute
newscasts each day simply cannot compete with a political ad that
runs several dozen times a day.
Limitations
Although every effort was made to obtain a random sample there is no
guarantee the three groups were truly representative. This would
present a threat to the external validity of the study. Another
threat to external validity exists in any laboratory environment that
cannot truly replicate a political campaign. The subjects viewed the
spots and news reports in an unrealistic setting. Nor were they
viewed in a real newscast. Although every effort was made to make
the experimental stimuli look real and natural, there is no question
the materials did not have the same visual and aural impact as they
would in a real campaign or news setting.
Lodge, Steenbergen and Brau (1995) remind us in their study, "The
Responsive Voter," that it is commonplace in laboratory settings for
subjects to be exposed to a message, be affected by that message, but
unable to recall that message. Simply put, human beings
"forget." Capella and Jamieson (1997) also observed that when
exposed to ad watches, subjects with lower levels of education were
able to pick up the tone of the ad watch, but could not recall many
details of the content.
This researcher contends that political ads, ad watches and race
profiles are complex. Many visuals, facts and claims are made inside
a very brief period of time. The viewer has much to absorb. The
structure and the content should both be considered confounding
variables for which it would be impossible to control completely.
Future Research
While McKinnon (1997) compared the effects of ad watches when viewed
on television and inside newspapers, future experimental research
could include ad watches viewed online. In general, more research is
needed on all forms of political news coverage online, and no doubt
this will arise from the study of the 2004 campaigns.
More research is needed to examine how television news departments
use political ads inside coverage that does not fall into the ad
watch category. If news producers are utilizing video and sound from
ads as general b-roll, what effects might that have on viewers'
opinions of the ads and the candidates? Experimental research might
determine whether using ads as b-roll could have a boomerang effect,
thus enhancing the message of the ad itself.
New research from the Lear Center indicates political campaign
coverage continues to decrease while political ad spending
increases. Follow-up research needs to determine why television
newsrooms are not investing in ad watch reports or any other kind of
political reporting.
Finally, since some of the results of this study are inconclusive,
replication of such studies might tell us more about which factors
can be reliably used to predict opinion of candidate, if any.
At best, conducting a study like this could help us move closer
toward building a better ad watch or race profile and understanding
the effects of such reports. If television station managers continue
to air more and more political ads during newscasts, those managers
must serve the public by providing ad watchdogs in the form of
electronic journalism. The answer seems not only to be "more"
political coverage, but also "better" coverage.
[1]
[2]
[3]
[4]
References
Ansolabehere, S., Iyengar, S. (1994). Riding the wave and claiming
ownership over issues: the joint effects of advertising and news
coverage in campaigns. Public Opinion Quarterly, 58,335-357.
Ansolabhere, S., Iyengar, S., Simon, A. & Valentino, N. (1994). Does
attack advertising demobilize the electorate? American Political
Science Review, 88(4) 829-838.
Ansolabehere, S., & Iyengar, S. (1995). Going negative: How
political advertisements shrink and polarize the electorate. New
York: Free Press.
Ansolabehere, S., & Iyengar, S. (1996). Can the press monitor
campaign advertising? An experimental study. The Harvard
International Journal of Press/Politics, 1(1), 72-86.
Ansolabehere, S., Iyengar, S. & Simon, A. (1999). Replicating
experiments using aggregate and survey data: the case of negative
advertising and turnout. American Political Science Review, 93(4), 901-907.
Atkin. C. & Heald, G. (1976). Effects of political
advertising. Public Opinion Quarterly, 40, 216-228.
Barker, E. (1942). Reflections on Government. New York: Oxford
University Press.
Bartels, L. (1996). Uninformed votes: Information effects in
presidential elections. American Journal of Political Science, 40(1), 194-231.
Basil, M., Schooler, C. & Reeves, B. (1991). Positive and negative
political advertising: Effectiveness of ads and perceptions of
candidates. In F. Biocca (Ed.), Television and Political Advertising
(Vol. 1, pp. 245-262). Hilldale, NJ: Lawrence Erlbaum.
Becker, L. & Doolittle, J. (1975). How repitition affects
evaluations of and information seeking about candidates. Journalism
Quarterly, 52, 611-617.
Bennett, C. (1997). Assessing the impact of ad watches on the
strategic decision-making process: A comparative analysis of ad
watches in the 1992 and 1996 presidential elections. American
Behavioral Scientist, 40(8), 1161-1182.
Berelson, B., Lazarsfeld, P., & McPhee, W.
(1954). Voting. Chicago: University of Chicago Press.
Campbell, A., Converse, P., Miller, W., Stokes, D. (1966). Elections
and the Political Order. New York: John Wiley & Sons, Inc.
Campbell, D. & Stanely, J. (1963). Experimental and
Quasi-Experimental Designs for Research. Chicaogo: Rand McNally.
Capella, J., & Jamieson, K. (1994). Broadcast adwatch effects: a
field experiment. Communication Research, 21 (3), 342-365.
Capella, J., & Jamieson, K. (1997). Setting the record
straight. Harvard International Journal of Press and Politics, 2 (1), 13-23.
Capella, J., & Jamieson, K. (1997). Spiral of Cynicism. New
York: Oxford University Press.
Chaffee, S., & Jamieson, K. (1994). Studies of the 1992 U.S.
election campaign. Communication Research, 21 (3), 261-263.
Devlin, L. (2001). Contrasts in presidential campaign commercials of
2000. American Behavioral Scientist, 44(12), 2338-2369.
Gobetz, R. & Chanslor, M. (1999). A content analysis of CNN "Inside
Politics" adwatch coverage of high-profile, nonpresidential
races. In L.L. Kaid & D.G. Bystrom (Eds.), The electronic
election: perspectives on the 1996 campaign communication
(pp.113-121). Mahwah, NJ: Erlbaum.
Goldstein, K., Freedman, P. (2002). Campaign advertising and voter
turnout: new evidence for a stimulation effect. Journal of Politics,
64(3), 721-740.
Hofstetter, C., Zukin, C. and Buss, T. (1978). Political imagery
and information in an age off television. Journalism Quarterly, 55, 562-569.
Hofstetter, C. & Buss, T. (1980). Politics and last-minute political
television. Western Political Quarterly, 33, 24-37.
Jamieson, K. H. (1992). Dirty politics: Deceptions, distraction, and
democracy. London: Oxford University Press.
Jamieson, K.H. (1999). Campaigns for Sale: A Newsroom Guide to
Political Advertising. Radio Television News Directors Foundation (a
publication of the Political Coverage Project, funded by the Carnegie
Foundation and Ford Foundation, project coordinator Avni Patel)
Jamieson, K.H. & Campbell, K. (1992). Interplay of Influence.
California: Wadsworth.
Jamieson, K. H., & Cappella, J. N. (1997). Setting the record
straight: Do adwatches help or hurt? Harvard International Journal of
Press/Politics, 2, 13-22.
Just, M., Crigler, A. & Wallach, L. (1990). Thirty seconds or
thirty minutes: What viewers learn from spot advertisements and
candidate debates. Journal of Communication, 40, 120-133.
Just, M., Crigler, A., Alger, D. , Cook, T. , Kern, M., & West,
D. (1996). Crosstalk: Citizens, candidates, and the media in a
presidential campaign. Chicago: University of Chicago Press.
Kaid, L. (1981). Political advertising.. In D.D. Nimmo & K.R. Sanders
(Eds.) Handbook of Political Communication (pp. 249-271). Beverly
Hills, CA: Sage.
Kaid, L.(2004). Political advertising. In L. Kaid (Ed.) A Handbook of
Political Communication Research. (pp. 155-188). Mahwah, New Jersey:
Lawrence Erlbaum Associates.
Kaid, L. & Boydston, J. (1987). An experimental study of the
effectiveness of negative political advertisements. Communication
Quarterly, Spring 1987, 193-201.
Kaid, L. & Chanslor, M. (1995). Changing candidate images: The
effects of television advertising. In K. Hacker (Ed.), Candidate
images in presidential election campaigns (pp. 83-97). New York: Pareger.
Kaid, L. & Johnston, A. (1991). Negative versus positive television
advertising in presidential campaigns, 1960-1988. Journal of
Communication, 41, 53-64.
Kaid, L., Chanslor, M., & Hovind, M. (1992). The influence of
program and commercial type on political advertising
effectiveness. Journal of Broadcasting and Electronic Media, 36, 303-320.
Kaid, L., Leland, C. & Whitney, S. (1992). The impact of televised
political ads: evoking viewer responses in the 1988 presidential
campaign. Southern Communication Journal, 57, 285-295.
Kaid, L., Gobetz, R., Garner, J., Leland, C., & Scott, D.
(1993). Television news and presidential campaigns: the
legitimization of televised political advertising. Social Science
Quarterly, 74 (2), 274-285.
Kaid, L. & Sanders, K.R., (1978). Political television
commercials: An experimental study of the type and
length. Communication Research, 5, 57-70.
Kaid, L. L., Tedesco, J. C. , & McKinnon, L. M. (1996). Presidential
ads as nightly news: A content analysis of 1988 and 1992 televised
adwatches. Journal of Broadcasting and Electronic Media, 40, 297-308.
Kaid, L., McKinney, M., Tedesco, J., & Gaddie, K.
(1999). Journalistic responsibility and political advertising: a
content analysis of coverage by state and local media. Communication
Studies, 50 (4), 279-293.
Kaid, L., McKinney, M., & Tedesco, J. (2000). Media
voices: journalistic assessment of candidate ads. In Civic
Dialogue in the 1996 Presidential Campaign (pp. 115-134). Cresskill,
NJ: Hampton.
Leshner, G. (2001). Critiquing the image. Communication Research,
28 (2), 181-208.
Lichter, S.R. & Noyes, R. (1996). Campaign '96: The media and the
candidates (first report to the Markle Foundation). Washington,
D.C.: Center for Media and Public Affairs.
Lodge, M., Steenbergen, M. & Brau, S. (1995). The responsive voter:
campaign information and the dynamics of candidate evaluation.
American Political Science Review, 89(2), 309-326.
McKinnon, L. M. (1995, May). Mediating political mudslinging or
magnifying advertising effects: an experimental study of adwatch
effects on voters' evaluations of candidates and their ads.
Unpublished doctoral dissertation completed at the University of Oklahoma.
McKinnon, L., Kaid, L., Murphy, J., & Acree, C. (1996). Policing
political ads: an analysis of five leading newspapers' responses to
1992 political advertisements. Journalism and Mass Communication
Quarterly, 73 (1), 66-76.
McKinnon, L. M., & Kaid, L. L. (1999). Exposing negative
campaigning or enhancing advertising effects: An experimental study
of adwatch effects on voters' evaluations of candidates and their
ads. Journal of Applied Communication Research, 27, 217-236.
Meadow, R. & Sigelman, R. (1982). Some effects and noneffects of
campaign commercials: an experimental study. Political Behavior, 4(2) 163-174.
Mills, L. (2003). An analysis of political ad coverage inside local
TV news during Florida's 2002 election reveals which stations did ad
watches and how they produced them. Paper presented at the meeting of
the Broadcast Education Association, Las Vegas, NV.
Mulder, R. (1979). The effects of televised political ads in the
1975 Chicago mayoral election. Journalism Quarterly, 56, 336-340.
Noggle, G. & Kaid, L. (2000). The effects of visual images in
political ads: experimental testing of distortions and visual
literacy. Social Science Quarterly, 81(4), 914-927.
O'Sullivan, P. & Gieger, S. (1995). Does the watchdog
bite? Newspaper ad watch articles and political attack
ads. Journalism and Mass Communication Quarterly, 72, 771-785.
Page, B. & Shapiro, R. (1992). The Rational Public : Fifty Years of
Trents in Americans' Policy Preferences. Chicago : University of Chicago Press.
Patterson, T. & McClure, R. (1976). The unseeing eye: Myth of
television power in politics. New York: Putnam
Pfau, M., & Louden, A. (1994). Effectiveness of adwatch formats in
deflecting political attack ads. Communication Research, 21 (3), 325-341.
Popkin, S. (1994). The Reasoning Voter: Communication and
Persuasion in Presidential Campaigns, 2nd ed. Chicago: University of
Chicago Press.
Richardson Jr., G.W. (1998). Building a better adwatch. Harvard
International Journal of Press and Politics, 3 (3), 76-96.
Roberts, M. & McCombs, M. (1994). Agenda-setting and political
advertising: Origins of the news agenda. Political Communication,
11, 249-262.
Schleuder, J., McCombs, M. & Wanta, W. (1991). Inside the
agenda-setting process: How political advertising and TV news prime
viewers to think about issues and candidates. In F. Biocca (Ed.),
Television and Political Advertising (Vol. 1, pp.
263-310). Hillsdale, NJ: Lawrence Erlbaum.
Shipler, D. (1986, April 15). Poll shows confusion on aid to Contras.
The New York Times, p. A6.
Tedesco, J. C., McKinnon, L. M., & Kaid, L. L. (1996). Advertising
watchdogs: A Content analysis of print and broadcast adwatches. The
Harvard International Journal of Press/Politics, 1(4), 76-93.
Tedesco, John C.; Kaid, Lynda Lee; and McKinnon, Lori Melton
(2000). Network Adwatches: Policing the 1996 Primary and General
Election Presidential Ads. Journal of Broadcasting and Electronic
Media, 44 (4).
West, D.M. (1993). Air wars: Television advertising in election
campaigns, 1952-1992. Washington, D.C.: Congressional Quarterly Inc.
Wattenberg, M. & Brians, C. (1999). Negative campaign
advertising: demobilizer or mobilizer? American Political Science
Review, 93(4) 891-897.
Zaller, J. (1992) The nature and origins of mass opinion. Cambridge,
UK: Cambridge University Press
Endnotes
1 Spending on political advertising reached another record high
during the 2004 presidential campaign, totaling about one-point-five
billion dollars. (Source: Campaign Finance Institute)
2 These results appear in news releases dated October 16, 2002 and
October 21, 2004 from the Norman Lear Center at the USC Annenberg
School in Los Angeles (www.learcenter.org). The Lear Center heads a
collaborative project funded by the Joyce Foundation and directed by
Professor Ken Goldstein at the University of Wisconsin-Madison.
3 These results appear in an article that appeared in Broadcasting &
Cable magazine dated November 18, 2002. Dan Trigoboff wrote the
article, but the research was conducted by Project for Excellence in
Journalism. The project is part of the Columbia University Graduate
School of Journalism and is underwritten by the Pew Charitable Trusts.
4 The survey is contained inside a report released at the April 2003
Radio Television News Directors Association Convention in Las
Vegas. The report, "2003 Local Television News Study of News
Directors and the American Public" was organized for RTNDA by the
Journalism Ethics Project, which is supported by the Ford Foundation.
|
