Effects of
Political Advertising
CHARLES ATKIN
AND GARY HEALD
OLITICAL candidates have relied increasingly on broadcast advertising
to inform and influence the electorate, but few research studies have
examined the impact of paid mass media messages on the voter. This investigation
explores the relationship of television and radio advertising
exposure to a variety of cognitive and affective variables in a typical con- *
gressional campaign. The research assesses how exposure variables relate
to (1) knowledge about the candiates and issues, (2) issue agenda priorities,
(3) interest in the campaign, (4) liking for each candidate, and (5)
polarized affect toward the candidates. Conditional relationships between
these variables are examined between subgroups of respondents
differing in initial familiarity with the candidates, exposure to other
sources, and motivation for advertising exposure.
Political knowledge is typically defined in terms of an individual's ability
to recall candidates' names, personal characteristics, and qualifications;
to identify election issues and current campaign developments;
and to recognize connections between candidates and issue positions.
The impact of general mass media campaign communications on gains in
knowledge has been inferred in numerous voting studies, based on recurrent
findings of a moderate association between media exposure and
campaign-related knowledge.1
1
Bernard Berelson, Paul Lazarsfeld, and William McPhee, Voting, Chicago, University
of Chicago Press, 1954; Joseph Trenaman and Denis McQuail, Television and the Political
Image, London, Methuen, 1961; Elihu Katzand J. J. Feldman, "The Debates in the Light
of Research: A Survey of Surveys," in Sidney Kraus, ed., The Great Debates, Bloomington,
Indiana University Press, 1962; Jay Blumler and Denis McQuail, Television in Politics, Chicago,
University of Chicago Press, 1969.
Abstract Relationships between broadcast advertising exposure and various cognitive
and affective orientations were assessed in a survey of voters during a congressional election
campaign. Exposure was moderately correlated with political knowledge and interest.
Highly exposed voters were somewhat more likely to attach higher agenda priorities to issues
and candidate attributes emphasized in the commercials. Personal affect toward each
candidate was mildly associated with advertising exposure frequency.
Charles Atkin is an Associate Professor in the Department of Communication at Michigan
State University. Gary Heald is an Assistant Professor of Mass Communication at
Florida State University. POQ 40 (1976) 216-228
byguestonFebruary5,2011poq.oxfordjournals.orgDownloadedfrom
EFFECTS OF POLITICAL ADVERTISING 2
' 7
Recent research evidence indicates that television advertising contributes
to voters' knowledge levels. McClure and Patterson report that
about three-fourths of the voters who recalled seeing a political advertisment
in the 1972 presidential campaign could correctly identify the ad's
message.2
Furthermore, voters heavily exposed to television were more
likely to show increased accuracy of perception of candidates' positions
on 10 issues presented frequently in campaign advertising: on the average,
there was a net 32 percent change in the correct direction among
heavy viewers and a net 24 percent change among light viewers.3
Atkin,
Bowen, Nayman, and Sheinkopf jfound that voters felt they learned substantive
information about candidates' qualifications and issue positions
from TV ads in two gubernatorial campaigns.4
The message and receiver conditions facilitating political knowledge
acquisition have been identified in several advertising studies. Patterson
and McClure discovered that political advertising had its strongest impact
on issue awareness for voters with low exposure to newspapers and
television news.5
Atkin et al. showed that perceived knowledge gain was
greatest for voters who paid close attention to advertising messages and
for those who cited an information-seeking motivation for watching ads.6
The role of message repetition in political knowledge gain has been
studied by Rothschild and Ray, who experimentally manipulated the frequency
of presentation of brief slide advertisments for several candidates.
They discovered that unaided recall increased monotonically from treatments
showing one to two to four to six repetitions; for instance, there
was 20 percent recall of the congressional candidates with a single presentation
and 55 percent recall with six presentations.7
2
Robert McClure and Thomas Patterson, "Television News and Political Advertising:
The Impact of Exposure on Voter Beliefs," Communication Research, Vol. 1,1974, pp. 3-31.
' Thomas Patterson and Robert McClure, "Television News and Televised Political Advertising:
Their Impact on the Voter," Congress and Mass Communications, appendix to
hearings before the Joint Committee on Congressional Operations, Ninety-third Congress,
Second Session, 1974, pp. 571-618.
4
Charles Atkin, Lawrence Bowen, Oguz Nayman, and Kenneth Sheinkopf, "Quality
Versus Quantity in Televised Political Ads," The Public Opinion Quarterly, Vol. 37, 1973,
pp. 209-224. In contrast, an unpublished survey of voters in nonpresidential campaigns
shows that those seeing the largest number of broadcast ads were the least capable of correctly
identifying the political candidates. This finding may be attributed to viewer selectivity
or a "clutter effect" that leaves the voter confused by excessive exposure to similar
campaign messages promoting minimally distinguishable candidates; see Timothy Meyer
and Thomas Donohue, "The New Student Voter and the 'Selling' of Politicians," Phi
Kappa Phi Journal, Vol. 54, 1974, pp. 8-13.
* Patterson and McClure, op. cit.
' Atkin et al., op. cit. To the extent that the political candidate's name and positions are
well known, exposure to campaign messages should not result in marked information gain.
In the 1960 presidential campaign, for example, there were increases in knowledge concerning
the relatively unknown Senator Kennedy but little changes in knowledge about the
more familiar Vice-President Nixon. See Ben-Zeev and I. S. White, "Effects and Implications,"
in Kraus, ed., op. cit.
' Michael Rothschild and Michael Ray, "Involvement and Political Advertising Effect:
An Exploratory Experiment," Communication Research. Vol. I, 1974, pp. 264-284.
byguestonFebruary5,2011poq.oxfordjournals.orgDownloadedfrom
2 1 8
ATKIN AND HEALD
The effect of political messages on the voters' agenda of priorities
among campaign issues and candidate attributes has periodically attracted
scholarly interest. Lazarsfeld, Berelson, and Gaudet observed
that the mass media had a marked effect in the 1940 presidential election
by redefining the issues such that "issues about which people had previously
thought very little, or had been little concerned, took on a new
importance as they were accented by campaign propaganda."8
Most of
the recent agenda-setting research has examined the relationship between
the rank-ordering of issues and attributes emphasized by the news media
and the perceived relative importance of these factors among the public.9
While the high correspondence between media and personal agendas may
be open to alternative explanations, most observers agree that to some
extent the media shape what people think about in arriving at a decision.
If a candidate can elevate the importance of those qualities and issues on
which he is positively perceived by most voters, the campaign may favorably
influence voters without actually persuading them to change issue
positions. The crucial goal may be to focus voter attention on which factors
to think about, rather than to convince them about what to think.
Considering agenda-setting and political advertising, Bowers compared
a content analysis of newspaper ads in a number of senatorial and
gubernatorial campaigns with Harris poll rankings of the importance of
campaign issues; an extremely high correlation was found between advertising
emphasis and voter emphasis.10
In a study of political TV advertising
and voter agenda in the 1972 presidential campaign, Shaw and Bowers
concluded that the appearance of an issue in commercials raised the
salience of that issue, especially for those persons exposed to the ads.11
These associations are undoubtedly due in part to candiates' reliance on
public opinion polling as a guide in deciding which issues to emphasize.
The role of the mass media in stimulating political interest in the electorate
is important because interested persons are more likely to vote in
an election.12
Interest is defined as the degree of concern or psychological
involvement in a particular election campaign. This variable has been
shown to correlate moderately with exposure to campaign content in
many studies.13
Lane proposes that increases in the availability of politi•
Paul Lazarsfeld, Bernard Berelson, and Hazel Gaudet, The People's Choice, New York,
Columbia University Press, 1948, p. 98.
9
Maxwell McCombs and Donald Shaw, "The Agenda-Setting Function of the Media,"
The Public Opinion Quarterly, Vol. 36, 1972, pp. 176-187; Jack McLeod, Lee Becker, and
James Byrnes, "Another Look at the Agenda-Setting Function of the Press," Communication
Research, Vol. I, 1974, pp. 131-167.
10
Thomas Bowers, "Newspaper Political Advertising and the Agenda-Setting Function,"
Journalism Quarterly, Vol. 50, 1973, pp. 552-556.
" Donald Shaw and Thomas Bowers, "Learning from Commercials: The Influence of TV
Advertising on the Voter Political 'Agenda,' " paper presented to the Association for Education
in Journalism, 1973.
12
Lester Milbrath, Political Participation. Chicago, Rand McNally, 1965.
" Lazarsfeld, Berelson, and Gaudet, op. cit.; Berelson, Lazarsfeld, McPhee, op. cit.
byguestonFebruary5,2011poq.oxfordjournals.orgDownloadedfrom
EFFECTS OF POLITICAL ADVERTISING 219
cal information will lead to increased politicization in a society.14
Atkin,
Galloway, and Nayman used cross-lagged correlational teachniques to
demonstrate that campaign interest and media exposure influence each
other—that interest produces exposure and exposure increases interest.15
Atkin et al. provide the only political advertising evidence on this question;
they discovered a mild positive relationship between attention to TV
commercials and campaign interest.16
Any association between these two
variables might be interpreted primarily as advertising effects, since one
of the basic goals of aggressive political advertising is to overcome the
barriers of voter apathy, and the unpredictable schedule of advertising
presentation reduces the opportunity for active information seeking by
those with prior interest.17
Liking for candidates is defined as a positive affective orientation toward
the candidate as a person, independent of party affiliation or issue
positions. Thus, liking is viewed as a personal attraction toward an individual
rather than an ideologically based evaluation. "Mere exposure"
theory suggests that repeated symbolic experience with a novel and
simple stimulus will lead to greater positive affect for the object por-
trayed.18
As applied to broadcast political advertising for little-known
candidates such as nonincumbents, there should be a positive logarithmic
correlation between the number of message exposures and degree of liking
for the candidate. This impact should be particularly strong for voters
with low involvement and low awareness of the candidate.19
Beyond this "mere exposure" effect may lie a more general polarization
of affective orientations toward candidates. Since even brief spot ads provide
substantive information along with projection of a name and face, it
is possible that exposure may produce counterproductive effects on voters
who are unimpressed or antipathetic to the candidate's attributes.
Perhaps a more likely outcome of repeated exposure is affective activation—the
formulation of a positive or negative personal evaluation of a
candidate and an increase in certainty of that orientation. Thus, con-
14
Robert Lane, Political Life, New York, Free Press, 1965.
15
Charles Atkin, John Galloway, and Oguz Nayman, "Reciprocal Causality Among Political
Interest, Political Knowledge and Mass Media Exposure," paper presented to the International
Communication Association, 1974.
16
Atkin, Bowen, Nayman, and Sheinkopf, op. cit.
" Robert Agranoff, The New Style in Election Campaigns. Boston, Holbrook Press, 1972;
Harold Mendelsohn and Irving Crespi, Polls, Television and the New Politicians. Scranton,
Pa., Chandler, 1970.
18
Robert Zajonc, "Attitudinal Effects of Mere Exposure," Journal of Personality and Social
Psychology Monograph Supplement. Vol. 9, 1968, pp. 1-27. This is rooted in the common
finding that attitudes people have toward one another are favorably enhanced by
social interaction; see Leon Festinger, "Group Attraction and Membership," Journal of
Social Issues. Vol. 7, 1951, pp. 152-163; George Homans, Social Behavior: Its Elementary
Forms. New York, Harcourt, Brace, 1961; Theodore Newcomb, "Stabilities Underlying
Changes in Interpersonal Attraction," Journal of Abnormal and Social Psychology. Vol. 66,
1963, pp. 376-386.
" Rothschild and Ray, op. cil.
byguestonFebruary5,2011poq.oxfordjournals.orgDownloadedfrom
2 2 0
ATKIN AND HEALD
sequences of intrusive advertising could include either liking or disliking,
depending on the predispositions of the voter. In a related finding, Atkin
et al. report that three-fifths of late-deciding voters were positively influenced
by the chosen candidate's ads, and fully half were unfavorably
affected by the unchosen candidate's ads.20
This investigation examines the associations between political advertising
exposure and the criterion variables of knowledge, agenda, interest,
liking, and polarized affect. Three advertising variables are studied: frequency
of exposure to TV ads, degree of attention to TV ads, and degree
of attention to radio ads. In addition, the research design includes a number
of third variables for purposes of comparison, control, and specification:
exposure to newspaper and television news, formal and informal interpersonal
communication, initial familiarity with the candidates, and
reasons for exposure to broadcast advertising.
Method
Interviews were conducted with a random telephone sample of 323
mid-Michigan voters during the last weeks of the 1974 congressional
campaign, focusing on a race between two candidates for the U.S. House
of Representatives. The candidates were nonincumbents competing in a
swing district, so each relied heavily on broadcast advertising to reach the
voters. The Democratic candidate was fairly well known from previous
campaigns, while the Republican candidate was almost a total unknown
before the campaign.
Prior to formulating the survey instrument, the campaign media directors
for each candidate were interviewed to determine which issues would
be emphasized in radio and television commercials. Based on their plans
and actual scripts, questions about issue knowledge and agenda emphases
were framed. The fifteen-minute survey asked questions on a number
of variables; following are brief descriptions of the variables of central
importance.21
PREDICTOR VARIABLES
Radio Attention: Attention paid to each candidate's ads while listening
to radio.
TV Exposure Frequency: Total number of ads for each candidate seen
during campaign.
" Similarly, voters who had decided on a candidate before the campaign began were just
as likely to say that opposition as well as their own candidate's advertising served to reinforce
their predispositions; see Atkin, Bowen, Nayman, and Sheinkopf, op. cit.
" The actual wording of the questionnaire items can be obtained by writing to the senior
author. Individual elements combined into the criterion variable indices are described at the
foot of Table I. Components of all indices are weighted equally; the advertising exposure index
is the sum of "radio attention" plus the product of "TV attention" times "TV exposure
frequency."
byguestonFebruary5,2011poq.oxfordjournals.orgDownloadedfrom
EFFECTS OF POLITICAL ADVERTISING Z 2 1
TV Attention: Closeness of attention to each candidate's ads on tele-
vision.
CRITERION VARIABLES
Knowledge: Correct recall of candidates' names and identification of issue
positions.
Agenda: Ratings of importance of certain issues and candidate attri-
butes.
Interest: Degree of concern about congressional campaign.
u Liking: Personal affect for each candidate, aside from issues and party
preferences.
i Polarized affect: Formation of positive or negative evaluation, and degree
of confidence in evaluation.
THIRD VARIABLES
News Exposure: Amount of newspaper article reading and TV news
story viewing relating to congressional campaign.
Interpersonal Communication: Frequency of talking about campaign
with family or friends, and contact with candidates or campaign workers.
Previous Familiarity: How well informed about each candidate before
campaign began.
Viewing Information-Seeking: Reason for viewing ads, either motivated
seeking of information or unmotivated exposure due to advertising
intrusiveness and entertainment seeking.
The data were analyzed to determine bivariate associations, partial
correlations controlling third variables, and conditional relationships at
two levels of several control variables. Since many of the "predictor" and
"criterion" variables are not clearly antecedent in a given relationship,
caution must be observed in inferring causality from advertising exposure.
In particular, the exposure-knowledge, exposure-interest, and exposure-affect
relationships may be due to reverse causation as the more
knowledgeable, interested, or favorable voters pay greater attention to
advertisements.
Results
The findings are presented separately for each of the five criterion vari-
ables.
KNOWLEDGE
The voters' knowledge of the candidates and their issue positions is
moderately correlated with radio and television advertising exposure;
Table 1 shows a correlation of + .42 for the overall index of broadcast
exposure. Learning of candidate names vs. issue stands occurs to an
byguestonFebruary5,2011poq.oxfordjournals.orgDownloadedfrom
221 ATKIN AND HEALD
Table 1. Communication Correlates of Knowledge, Agenda, Interest, and Polarization*
Exposure Variable
Advertising exposure index
TV ads frequency
Attention on TV ads
Attention of radio ads
News exposure index
Newspaper article frequency
TV news story frequency
Interpersonal exposure index
Conversation frequency
Candidate/worker contact
Advertising exposure index
News exposure index
Interpersonal exposure index
(Advertising, news, interpersonal)
r
Knowledge
+ .42
+ .34
+ .34
+ .27
+ .39
+ .43
+ .20
+ .36
+ .29
+ .29
+ .26
+ .20
+ .18
.51
between Exposure and
Agenda
+ .18
+ .16
+ .15
+ .13
+ .18
+ .18
+ .12
+ .07
+ .08
+ .03
Interest
+ .40
+ .28
+ .38
+ .30
+ .42
+ .36
+ .31
+ .41
+ .38
+ .25
Beta weights
+ .13 +.20
+ .14 +.22
-.03 +.24
Multiple correlations
.22 .52
Polarization
+ .28
+ .19
+ .28
+ .19
+ .31
+ .32
+ .19
+ .29
+ .25
+ .21
+ .13
+ .18
+ .16
.38
* Knowledge Index = (recalls Democratic candidate's name + recalls Republican
candidate's name + knows Democrat's position on military spending + knows Republican's
position on balancing budget + knows Republican's position on being own man).
Agenda Index = (nominates military spending as important issue + nominates
balancing budget as important issue + degree of importance of candidate's position on
military spending + degree of importance of candidate's position on balancing budget +
degree of importance for candidate to be own man + degree of importance for candidate
to be sensitive to people's needs).
Interest Measure = degree of interest in congressional election.
Polarization Index = (degree of liking or disliking for Democrat + degree of liking
or disliking for Republican).
Exposure indices are composed of individual items listed under each index.
equivalent degree (not shown). For comparative purposes, political
knowledge correlates +.39 with the index of newspaper and TV news
campaign exposure, and +.36 with the interpersonal exposure index.
These factors combine to produce a multiple correlation of .51.
The advertising-knowledge relationship is reduced to a partial correlation
of +.25 with exposure to these other sources controlled. Additional
partialling on precampaign familiarity with the candidates and campaign
interest further reduces the association to +.21. Thus, with likely contaminating
factors controlled, a substantial relationship remains between
broadcast advertising exposure and campaign knowledge, indicating a
functional relationship. The finding that frequency of viewing commercials
is correlated +.34 with knowledge provides evidence that advertising
exerts a causal influence in the relationship, since sheer exposure
byguestonFebruary5,2011poq.oxfordjournals.orgDownloadedfrom
EFFECTS OF POLITICAL ADVERTISING 2 2 3
frequency is unlikely to be due to selective attention by the knowl-
edgeable.
Table 3 shows that the relationship is stronger among respondents who
reported an informational motivation for viewing advertising. Those who
said they watched because "the ads came on" or solely because ads were
entertaining appear to learn less information than the explicit information-seekers.
On the other hand, there is no differential association at
high vs. low levels of previous familiarity or exposure to nonadvertising
messages.
AGENDA
Voters were asked an open-ended question about the issues that they
personally considered most important in making their congressional voting
decision, and this was followed by several items asking for rating of
the importance of the candidates' positions on several issues and the importance
of several candidate qualities. Some issues and attributes were
emphasized by both candidates, some by one candidate, and some by neither
candidate.22
Table 1 shows that the overall agenda index is associated
+ .18 with the three advertising exposure items, along with the
news media exposure index. Interpersonal exposure is only slightly re-
lated.
Since the agenda-setting findings are the unique feature of this study,
these data are presented in more explicit detail in Table 2. The sample
was divided at the median on the advertising exposure index, and responses
were cross-tabulated on each agenda item. Those who were highly exposed
are somewhat more likely to name as important the two issues
most heavily emphasized in the advertising campaign, although the differences
are not significant. On the four emphasized issues specifically
cited in close-ended questioning, there is a consistent but modest tendency
for the highly exposed respondents to feel that these factors are
more important than do the less exposed respondents. The only significant
finding is the 94 percent vs. 82 percent difference between these two
groups in rating that it is "very important" for the candidate to be sensitive
to people's needs. There is also a consistent tendency for less exposed
voters to reply that they "don't know" the importance of the four issues.
Two approaches are used to test the validity of the relationship. Partialling
on news and interpersonal exposure, previous familiarity, and
campaign interest reduces the correlation somewhat from +.18 to +.12.
Furthermore, a second index was constructed of issues not emphasized in
" A four-item agenda index was constructed for each candidate based on the extent to
which they gave priority to each factor in their TV and radio advertising. The correlation
between exposure to the Republican's advertising and the voters' own agenda was +.18;
and identical +.18 correlation was obtained for Democrat ad exposure and the voters'
agenda. Since the two exposure indices and two agenda indices are strongly correlated and
the relationships are quite similar, an overall agenda index was pooled for further analyses.
byguestonFebruary5,2011poq.oxfordjournals.orgDownloadedfrom
224 ATKIN AND HEALD
Table 2. Cross-Tabulations between Exposure to Broadcast Advertising Exposure and
Importance of Campaign Issues
Importance Item
There are a number of issues and problems facing
the voters in this campaign for Congress. In
making your voting decision, which of the
issues do you personally consider most im-
portant?
Mentioned balancing the federal budget
Mentioned cutting military spending
How important is the candidate's position on
balancing the federal budget?
Very important
Somewhat important
Not so important
Don't know
How important is the candidate's position on cutting
military spending?
Very important
Somewhat important
Not so important
Don't know
How important do you feel it is for the candidate to
be his own man?
Very important
Somewhat important
Not so important
Don't know
How important do you feel it is for the candidate
to be sensitive to the needs of the people?
Very important
Somewhat important
Not so important
Don't know
Amount ofExposure
Low
N = 165
15%
7%
60%
21
13
6
52%
27
15
6
62%
16
11
11
82%
11
3
4
High
N= 158
23% XJ
= 2.6
11% X 2
= 1.7
67% X1
= 4.0
21
10
2
51% X2
= 3.6
35
10
4
67% X2
= 6.0
16
13
4
0> = 10)
(p = .19)
(p = .26)
0* = -30)
0 = 1 1 )
94% X ' = 11.5 0* = .01)
5
1
0
the advertising (importance of Nixon pardon, forced busing, energy
crisis, cost of living, unemployment). This nonagenda index was correlated
-.02 with broadcast advertising exposure. Thus, the issues given
priority in the commercials were perceived by exposed voters as important,
while no difference by exposure was found for the "control" issues.
This suggests that the relationship between the candidate and voter
agendas is functional (although not strong); the direction of causality is
likely to be from advertising to agenda.
The conditional associations in Table 3 show that the relationship is
substantially greater among respondents not previously familiar with the
candidates and among those not highly exposed to other information
sources. No difference occurs according to the level of motivation for exposure.
Thus, the minimally informed voters appear to be most strongly
byguestonFebruary5,2011poq.oxfordjournals.orgDownloadedfrom
EFFECTS OF POLITICAL ADVERTISING 2*5
Table 3. Conditional Correlations between Political Advertising Exposure and Knowledge,
Agenda, Interest, Polarization, and Liking
Conditional Variable
Pre-campaign familiarity with
candidates
Not informed (N = 137)
Informed (N = 186)
News and interpersonal exposure
Below median (JV = 149)
Above median (N = 174)
Motivation for advertising viewing
Information (yv = 96)"
Other/none (N = 84)
r between Advertising Exposure Index and
Knowldge
+ .36
+ .37
+ .28
+ .33
+ .43
+ .30
Agenda
+ .25
+ .15
+ .24
+ .09
+ .10
+ .10
Interest
+ .41
+ .29
+ .18
+ .33
+ .42
+ .38
Polari-
zation
+ .24
+ .23
+ .27
+ .16
+ .12
+ .04
Liking*
+ .18
+ .05
+ .21
+ .08
+ .11
+ .05
• Figures in the Liking column are the average of the correlation between exposure
and liking for the Democratic candidate and the exposure-liking correlation for the
Republican.
0
/V's in the Motivation conditions are reduced because only those voters who had
seen both candidates' commercials were analyzed.
influenced by the agenda set in the candidates' advertising. Those more
cognitively involved in the campaign are less susceptible to agenda-setting
effects.
INTEREST
Advertising exposure is correlated +.40 with the campaign interest
measure; news and interpersonal exposure are related to a similar degree.
The multiple correlation of these three factors and interest is .52.
Since the raw association is likely to be highly spurious, previous familiarity,
knowledge, and exposure to news and interpersonal messages were
controlled. The partial correlation is a modest +.15, with the news and
interpersonal exposure index accounting for most of the reduction.
The conditional relationship between advertising exposure and interest
is stronger for voters with above average exposure to other communications
than among those less exposed to nonadvertising sources (Table 3).
There is a higher correlation among voters who were not informed prior
to the campaign than among those who knew something about the candidates.
No difference occurs by motivation for viewing.
LIKING
Advertising exposure measures are consistently related to positive affect
toward each candidate, as shown in Table 4. An index of exposure to
the Democrat's broadcast ads is correlated +.16 with liking for this candidate,
and the Republican's exposure index is correlated +.20 with lik-
byguestonFebruary5,2011poq.oxfordjournals.orgDownloadedfrom
Liking/or
Democrat
+ .16
+ .11
+ .16
+.11
+ .08
+ .09
+ .08
+ .03
+ .09
+ .12
+ .11
+ .07
Likingfor
Republican
+.07
+ .01
+ .10
+.01
+ .20
+ .13
+ .20
+ .11
+ .13
+ .10
+ .19
+ .08
M ATKIN AND HEALD
Table 4. Political Advertising Correlations of Liking for Congressional Candidates
r between Exposure and
Exposure Variable
Democrat advertising exposure index
Number of Democrat's TV ads viewed
Degree of attention to Democrat's TV ads
Degree of attention to Democrat's radio ads
Republican advertising exposure index
Number of Republican's TV ads viewed
Degree of attention to Republican's TV ads
Degree of attention to Republican's radio ads
Partial correlations
Democrat Advertising Exposure Index
controlling news and interpersonal exposure
controlling campaign interest
controlling previous familiarity with Democrat
controlling for all three variables
Republican Advertising Exposure Index
controlling news and interpersonal exposure
controlling campaign interest
controlling previous familiarity with Republican
controlling for all three variables
ing him; the correlations with specific types of advertising exposure range
from +.11 to +.20.23
Two approaches are used to determine the substance of these correlations.
First, correlations of exposure to a candidate's ads with personal
affect toward that candidate are compared to correlations with affect toward
the opposing candidate; in all cases, advertising for a candidate is
more closely related to affect toward him rather than his opponent. Second,
partialling on interest, precampaign familiarity, and other communication
exposure yields diminished yet positive associations. Thus, there
is consistent evidence that exposure is related to liking, although the magnitude
of the relationship is marginal.
Conditional analyses in Table 3 show that the exposure-affect relationship
exists primarily among those who had no prior familiarity; those
23
The specific role of mere exposure was examined with correlations based on the frequency
of TV commercial viewing; the median number of ads viewed for each candidate
was four, while the mean was about seven. The number of times that the respondent reported
seeing ads for the Republican candidate was correlated +.13 with liking him; the
corresponding Democrat's correlation was +.11. Zajonc suggests that message repetition
should be defined as a logarithmic rather than a direct linear function; when the log of the
exposure frequency variable was used in analysis, the correlation increases minimally to
+ .14 for the Republican while the Democrat's correlation remains at +.11. Both drop to
+ .04 when the control variables are applied. It appears that the development of positive affect
through repeated exposure occurs only to a slight degree; indeed, this minimal relationship
might be due to favorable voters watching (or remembering watching) more ads.
byguestonFebruary5,2011poq.oxfordjournals.orgDownloadedfrom
EFFECTS OF POLITICAL ADVERTISING J 2 7
who were aware of a candidate before the campaign were not influenced
by repeated exposure, while the uninformed group has a correlation of
+ .26 in the case of the Republican candidate and +.10 in the case of the
Democrat. Furthermore, the relationship is considerably stronger for respondents
with lower exposure to other information sources. These conditional
relationships are consistent with "mere exposure" theory. Finally,
the motivated viewer group has a somewhat higher association
than the group not seeking information from ads.
POLARIZATION
Ratings along the personal affect scale were transformed so that "like
very much" or "dislike very much" received the highest score while "neutral"
or "don't know" was scored lowest; the degree of affect was then
summed across the two candidates. This index of polarization is correlated
to the same moderate extent with advertising, news, and interpersonal
exposure (Table 1). Partialling on the control variables, the relationship
between advertising exposure and polarization drops sharply
from +.28 to +.12. Apparently a functional association remains, but the
magnitude is modest when these factors are considered. Table 3 shows
that the voters less exposed to nonadvertising messages and more motivated
to obtain information have the larger correlations between advertising
exposure and polarization.
Discussion
The pattern offindingsshows that exposure to radio and television advertising
is positively related to all criterion variables, although the generally
moderate zero-order correlations decline substantially when other
predictor variables are controlled. The evidence indicates that advertising
exposure is functionally related to knowledge, agenda, interest, affect,
and polarization. Whether exposure is a cause or consequence of these
variables is difficult to determine in a single-shot survey; however, it appears
unlikely that selective seeking of advertising messages could fully
account for the relationships, since reception of broadcast commercials is
often due to chance opportunity or to entertainment seeking. Thus, it
seems that advertising does contribute to the voters' cognitive and affective
orientations to some extent.
Considering the various communication predictor variables, the results
indicate that advertising exposure, news exposure, and interpersonal exposure
are related to almost equal extents with the political orientations
of the voters/The main exception is the weak relationship between interpersonal
communication and agenda priorities. Exposure to newspaper
articles about the congressional campaign is the strongest single predictor
of the criterion variables. Among the advertising predictors, attention to
byguestonFebruary5,2011poq.oxfordjournals.orgDownloadedfrom
M8 ATKIN AND HEALD
TV ads is the strongest correlate in almost every case; radio attention and
number of TV ads viewed are much less closely related to the political
orientation variables. The exposure variables requiring the most active
participation (TV attention and newspaper reading) are those most
strongly associated with the criterion measures.
Examination of the antecedent and intervening conditions for advertising
learning yields some interesting results. Somewhat stronger correlations
are found for those voters who were least well informed about the
candidates before the campaign began; for several criterion variables,
those less attentive to other campaign communication show stronger
relationships. Thus, it appears that voters who have few prior or concurrent
informational inputs may learn the most from advertising. Voters
who are manifestly using broadcast advertising for informational purposes
consistently show stronger correlations between exposure and orientation,
indicating that the motivation for attending to advertising is an
important factor.
In conclusion, the evidence suggests that a well-designed and well-financed
political advertising campaign in the broadcast media can serve
to (1) increase the electorate's level of knowledge about the candidate
and his featured issue positions, (2) elevate emphasized issues and attributes
higher on the voters' agenda of decisional criteria, (3) stimulate the
electorate's interest in the campaign, (4) produce more positive affect toward
the candidate as a person, and (5) intensify polarization of evaluations
of the candidate. It must be recognized that these consequences
of advertising occur within the context of other mass media and interpersonal
message inputs, and that the political orientations held or
learned by the voter probably exert a reciprocal influence in producing
exposure to broadcast advertising.
byguestonFebruary5,2011poq.oxfordjournals.orgDownloadedfrom