Greenberg 1965

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

VOTING INTENTIONS, ELECTION EXPECTATIONS

AND EXPOSURE TO CAMPAIGN INFORMATION*


S. GREENBERG*
BRADLEY *
A common, critical variable in both the major studies of Amer-
ican voting behavior (1,3,4,8) and the many recent investigations
of cognitive consistency (2,9,12) is selective exposure to informa-
tion-both to mass media and to interpersonal information
sources. The voting studies focus on active avoidance of com-
munication from the opposition. The principal notion among the
balance theorists is that an apparent inconsistency in important
beliefs can be resolved by exposure to reinforcing information
and avoidance of discrepant information. These two research
interests converge in the present study, which explored the use
of election information sources among voters whose expectations
regarding the outcome of a pending, local election were either
consistent or inconsistent with their own voting intenti0n.l
The importance of this problem is concerned with the function
of information in the election process. For many local elections,
disinterest is high. Also there is a high proportion of voters who
are inconsistent between what they want and what they think
is going to happen. The dissemination of information is pre-
sumed to contribute to understanding, and understanding to
public support of public issues. Hence, to the extent that cam-
paign communications fulfill a particular social-psychological
need and to the extent that we can specify the manner in which
that function is performed, we increase our understanding of
voting behavior.
This study focuses on the interrelationships among four vari-
able classes: ( a ) The antecedent conditions which predispose an

* The data reported in this paper are from a study conducted as part of
a project (No. 1039) under contract with the Cooperative Research Pro-
gram of the U. S. Office of Education at the Institute for Communication
Research, Stanford University. The project director was Prof. Richard F.
Carter, now at the University of Wisconsin.
* * Bradley S. Greenberg is Assistant Professor of Communication at
Michigan State University.
An earlier application of consistency theory concepts to communication
behavior in an election context is reported by R. Carter. [ 5 ]
149
150 Journal of Communication

individual to be interested in an election; ( b ) the decision to


vote one way or the other; ( c ) the individual’s perception that
he is voting with the winning or the losing side; and ( d ) the
magnitude of exposure to campaign communications. The em-
pirical relationships among this set of variables may be summa-
rized as follows: [ 11
a. Election interest is positively correlated with amount of exposure
to campaign information, the decision to vote (and making that decision
early in the campaign), and the expectation that the chosen candidate
will win.
b. The decided voters are the more exposed voters.
c. There is substantial expectation that one’s own side will be the win-
ning one.

One aspect of this configuration is unresolved-the relation-


ship between an individual’s expectations about the outcome of
the election and his exposure to campaign communications. Con-
ceivably, this aspect contains a segment of voters whose beliefs
about the election are internally inconsistent. These would be
the voters who have committed themselves to supporting some
side, but who perceive that “My side will lose.” They contrast
with the voters who are equally committed to one side, but who
believe it is the winning side. Such a division of committed
voters would make it possible to examine the relationship of
election expectations to information exposure. (The uncom-
mitted voter is less interested, attaches less importance to this
belief area, and therefore is omitted from the present conceptual-
ization.) In contrast with the relatively few individuals in na-
tional election studies who believed their candidate would lose,
fully one-third of the decided and committed voters in the local
elections described below believed that their side would lose out-
right or was as likely to lose as to win.
Pertinent evidence from two sources suggests that the voter
with consistent cognitions exposes himself to more different
sources of information than the voter whose intended vote and
election prediction are discrepant. First are the positive rela-
tionships among all other variable pairs in this cluster, which
suggest that the relationship between consistent cognitions and
exposure is likely to be positive. Second, the specific prediction
of a positive relationship stems from the demonstrated function
Exposure To Campaign Information 151

of mass and interpersonal communication channels to reinforce


existing attitudes and beliefs. [lo] For those voters who per-
ceive that they are voting with the majority, selective perception
and retention of campaign information will tend to be reinforc-
ing. The same information available to voters who think they
are backing a loser will be discrepant with one or the other of
these two cognitions and consequently punishing, rather than
reinforcing. Therefore, interested voters with inconsistent cog-
nitions are expected to be less exposed to campaign information
than equally interested voters with consistent ideas about their
vote and the likely winner.

METHOD
These data were gathered as part of a series of studies of per-
son-to-person communication about local school issues. [71 Field
studies were conducted in five school districts where a school
bond or tax election was pending within 10 days of the first in-
terviews. In each district, a probability sample of 50 house-
holds was selected. All persons 21 or over in each household
were interviewed, a total of 468 interviews. This was 85 per-
cent of the total adults in the sample households. Of these
people, 23% had recently talked about the local schools with
someone other than a family member. Interviews then were
conducted with persons to whom someone in the original sam-
ples ( 0 s ) had talked-or to someone who talked with some-
one in the original samples, etc. All respondents who were not
members of the original samples are designated as referrals ( R ) ,
and are examined separately. Interviews were completed
through two orders of referrals in each district and some inter-
viewing was completed with third- and fourth-order referrals.
The total number of referral interviews completed in this study
was 849. This represented 90 percent of all the referrals identi-
fied.
The variables were:
Information exposure. Respondents were asked whether they
had recently been exposed to each of seven sources of campaign
information: Radio, television, newspapers, public meetings and
speeches, club meetings and social gatherings, school board
meetings, and bulletins and pamphlets. An information exposure
152 Journal of Communication

score was obtained by summing each respondent’s affirmative


responses for questions concerning exposure to the seven infor-
mation sources (range = 0-7). The analyses examined both the
composite exposure level and the use of particular media.
Vote intention. The voting question was, “If the election were
being held today, how do you think you would vote on (the
request for an increase in the school tax or the school bond
issue) ?” Voters were classified as “for” or “against” the measure.
Undecided voters are omitted from the analyses.
Election expectation. The following question probed the re-
spondents’ perception as to the outcome of the election, “Do
you think the proposal will pass or fail?”
From responses to this item and the vote intention, some re-
spondents were classified as having consistent cognitions-those
who favored the measure and thought it would pass plus those
who opposed the measure and thought it would fail. Of the
original sample respondents, 69% were classified in this manner;
of the referrals, 77% were so classified. The remaining respon-
dents with a specific vote intention were classified as having in-
consistent cognitions-there was a discrepancy between their
vote intention and their perception of the outcome.2
Background characteristics. Certain background characteris-
tics predispose eligible voters to go to the polls or stay at home
in school district financial elections, and to vote in favor of or
against the school’s request. [4] In order that the backgrounds
of the consistent and inconsistent voters might be compared,
data were gathered on four aspects of demography (sex, age,
education, and size of family), and three aspects of participation
in civic and school affairs (membership in parent-school orga-
nizations, membership in civic groups, and perceived participa-
tion in school activities). In addition, we measured the respon-
dents’ perceived interest in school affairs, their evaluation of the
local schools, their pride in the local schools, and their perceived
efficacy of participation in local school affair^.^
Of those who favored the measure, 78% thought it would pass; of those
who opposed it, 48% thought it would fail. This greater consistency among
the favorable group is controlled for in the analyses, as explained in the
results section.
3Each of these final measures was a Guttman Scale. The i t e m com-
prising the scales are reported in R. Carter. [4]
Exposure To Campaign Information 153
TABLE1
Mean Information Exposure Scores For Consistent and Inconsistent Voters
Among Original Sampk (0s)and. Referral ( R ) Respondents,
b y School District
School Consistent Inconsistent F* df P
district cognitions cognitions

0s: 1.24 0.96


A 14.2 1/277 <O.OOl
R: 2.63 2.54
0s : 1.18 1.22
B 3.5 1/213 <0.10
R: 3.11 2.45
0s: 1.92 1.22
C 17.6 1/210 <O.OOl
R: 3.43 2.71
0s : 1.44 1.06
D 8.3 1/263 <O.OOl
R: 3.24 2.69
0s : 2.10 1.90
E 5.5 1/239 <0.02
R: 3.81 3.48
~ ~

* For the data in each district, an analysis of variance was performed. The F pre-
sented is that obtained between the Consistent and Inconsistent voter groups. None of
the interactions was significant.

RESULTS
Media Exposure
General exposure to sources of information for the voter
groups is displayed in Table 1. The findings within each of the
five study areas for both original sample and referral respondents
are remarkably alike.4 Only in District B was there no difference
in the media usage of the two voter groups within the original
sample. Even there, the marked difference among the referrals
resulted in an overall difference in the predicted direction ( F =
3.5, P < 0.10). In each of the other districts, high statistical sig-
nificance was achieved-the voters who thought their side would
win exposed themselves to more sources of information than the
voters who thought their side would lose.
4 T h e original sample and referral groups were examined separately to
control for possible interactions between media exposure, cognitive con-
sistency, and the original sample-referral classifications. No significant in-
teractions were obtained in any of these analyses.
154 Journal of Communication

Separate analyses by direction of vote does not alter this prin-


cipal finding. Among the favorable voters in all five districts,
the mean number of information sources used by those with
consistent beliefs was significantly larger than the average among
those with inconsistent cognitions ( F = 16.4, df = 1/1042, P <
0.001). Among the opponents of the financial measures, the dif-
ferences were in the same direction, but the total number of
declared negative voters was too small to permit extended analy-
sis. Lacking sufficient negative voters for separate analyses and
because other factors distinguish between those who favor and
those who oppose school financial measures, [4]all subsequent
analyses are restricted to those voters who intended to support
the financial proposals. The consistent ones believed the mea-
sure would pass; the inconsistent ones expected otherwise. Hence
any obtained differences cannot be due to different voting in-
tentions.
Here, we have clearly obtained a significant relationship be-
tween cognitive consistency and exposure to multiple sources
of campaign information, whether we look at all voters or at
the preponderant group of favorably inclined voters.

Exposure to Specific Sources of Campaign Communications


The next question was whether the greater exposure to sources
of campaign information among the consistent voters was gen-
erally diffused among several sources---or oriented to some few
sources. If, as reasoned, selective exposure to reinforcing infor-
mation was in part responsible for greater consistency, then a
small number of information sources would have performed that
function, principally the media with a more specific election
orientation. Individuals could more selectively choose meetings
to attend or pamphlets to read than radio or television content.
This expectation was partly confirmed by the reported relative
exposure rates of the voter groups to the seven different informa-
tion sources (see Table 2). The use of five information sources
was quite similar for both groups. There were no significant
differences in the use of television, radio, attendance at school
board meetings, clubs or social gatherings, or public speeches.
The principal difference was in attention to pamphlets, bulletins,
and other print information. Much of this additional informa-
Exposure To Campaign Information 155
TABLE2
Proportions of Consistent and Inconsistent Voters Exposed
to Each of Smen Sources of Campaign Information

Consistent Inconsistent
Information cognitions cognitions
sources ~

( n = 236) (n=78)

Television 11% 13%


Radio 12% 8%
Newspapers 51% 42%
School board meetings 3% 3%
Clubs, social gatherings 17% 13%
Public speeches 8% 4%
Bulletins, pamphlets 40% 27%

tion originated with the schools themselves or with citizens’


groups promoting the school’s financial request. A significantly
greater proportion of the consistent voter group had been ex-
posed to information from that single source ( t = 2.2, P < 0.05).
There was also a tendency for the consistent voters to have read
more about the election in their newspapers. The same pattern
of media emphasis between the consistent and inconsistent voter
groups was obtained separately for the referral respondents.

Who Holds Consistent OT Inconsistent Cognitions?


It was essential to determine if there were demographic, at-
titudinal, or other differences between the consistent and in-
consistent voter groups which in turn might account for the ob-
served differences in media exposure. If such variables were
located, it was necessary to control for them in examining the
positive correlation between cognitive consistency and media
exposure. For example, interest in political affairs is related to
media exposure and the expectation that one’s own candidate
will win. Therefore, it was necessary to determine separately
whether media exposure were related to cognitive consistency
both for those claiming considerable interest and for those with
less interest in school affairs.
The consistent and inconsistent voter groups among the orig-
inal sample respondents were compared on the 11 background
characteristics listed earlier. These comparisons were not made
156 Journal of Communication
TABLE3
Family Size, Participation, and Attitudes of
Consistent and Inconsistent Voters

Consistent Inconsistent (n)


cognitions cognitions

Number 0 64% 36%


of children 1 75% 25%
>1 79% 21%
PTA No 71% 29%
membership Yes 82% 18%
Civic group None 70% 30%
membership Some 78% 22%
Direct
participation Low 69% 31%
in school High 83% 17%
activities
Pride in the None m% 32%
local schools Some 79% 21%
Evaluation of Poor 67% 33%
the local Fair 74% 26%
schools Good 81% 19%
Communicated LOW 61% 39%
interest in the Moderate 73% 27%
local schools High 85% 15%
Rows sum to 100%.

among the referral groups (who were not a population sample,


but a clustering of informal communicators). On four of the
background characteristics, no significant differences were found
(on sex, age, education, and perceived efficacy of participation
in school affairs).
Appreciable differences were obtained for the remaining back-
ground variable (see Table 3 ) : ( 1 ) In terms of family size,
a larger proportion of the non-parents were in the inconsistent
voter group. ( 2 ) In terms of gregariousness, more of the inac-
tive citizens held inconsistent beliefs. ( 3 ) More PTA members
than nonmembers expected their side to win; ( 4 ) members of
community organizations reacted the same way; ( 5 ) more of
those who participated directly in school affairs and who com-
municated interest in the schools had consistent cognitions. Fi-
Exposure T o Campaign Information 157
TABLE4
Mean Information Exposure Scores For Consistent and
Inconsistent Votem by Background Characteristics

Background Consistent Inconsistent


characteristics cognitions cognitions

Parents 1.47 1.30


Non-parents 1.34 1.00
PTA members 1.90 2.15
Nonmembers 1.40 0.89
Civic group members 1.76 1.36
Nonmembers 1.27 1.00
High participation 2.10 2.09
Low participation 1.46 1.04
Some pride 1.82 1.61
No pride 1.10 0.72
High evaluation 1.72 1.33
Low evaluation 1.40 1.07
High interest 2.16 2.14
Low interest 1.44 1.05

nally, both variables describing school-related attitudes differ-


entiated the consistent from the inconsistent voter: ( 6 ) More of
those who reported some object of pride in the local schools and
( 7 ) who highly evaluated the local schools maintained consistent
cognitions.
In summary, the voter with consistent cognitions was a differ-
ent person, socially and psychologically, from the voter with
inconsistent cognitions. He was more likely a parent, an active
member of both school and civic organizations, with pride and a
favorable attitude toward the local schools, who communicated
that high degree of interest to others, And these consistent dif-
ferences were obtained among the samples of citizens all of
whom intended to vote in favor of the financial proposal.
Information exposure scores were then computed separately
for each level of the seven background attributes which differ-
entiated the voter groups (see Table 4). Controlling for each
of the related background characteristics, the predicted differ-
ence in media exposure between the voter groups was found in
13 of 14 possible comparisons. There was a tendency for the
158 Journal of Communication

maximum relationship between exposure score and consistency


to occur within that level of each background variable which
was more characteristic of inconsistent beliefs than consistent
ones. To illustrate, the difference in exposure scores between
the consistent and inconsistent non-parents is larger than be-
tween the parent groups ( Table 4). The discrepancy in informa-
tion sources between the voter types was also greater among
those without specific pride in the local schools, among non-
members of school groups, among weak participants in school
activities, and among those who lacked interest in school affairs.
We can offer a plausible explanation for this latter tendency.
Here, we have shown that voters of inconsistent beliefs with
background characteristics which should have predisposed them
to consistency, behaved more like those with consistent cogni-
tions in terms of exposure to campaign communications. The
former were committed to supporting one side in the election-
the normative pro side. They were “involved in the campaign-
in terms of interest and participation in civic and school activities
as well as having a vested interest in the future of their children.
The election was an important concern. For them, the incon-
sistency between their vote and their expectation should have
been least tolerable. Some effort should have been exerted to
reduce that inconsistency. Attempts to make the beliefs con-
sistent-by changing either the vote or the expectation-reason-
ably could have led to seeking additional information (6,ll).
This then would account for the greater media-directed behavior
on the part of this subgroup of inconsistent voters, in contrast
with the inconsistent voters whose backgrounds predisposed
them to be less motivated regarding the election and its outcome.

DISCUSSION
Regardless of how the voters were stratified-by direction of
vote, by background characteristics, by origination in original
sample or referral groupings-those who thought the side they
supported would win had been exposed to significantly more
campaign information than those who thought their side would
not win. More of the voters with consistent cognitions used
bulletins, leaflets, and other print matter which could more
easily be screened by selective processes.
Exposure To Campaign Information 159

FIGURE 1
Postulated Voter Paradigm of Cognitive Consistency,
lnformtwn-Seeking, and Media Exposure

TI _ - - _- _ _ _ _ _ _ _ _ _ _ T,
_ _ _ _ _ _ _ _ _ T, _ _ _ - - _ _ _ _ T,
1 1 1
Voter Group A: C C C
---4--- - - -3.5-- - ---3---
3 2 2.5
Voter Group B: I C C
---I--- ---3.5 - - - ---3---

3 3.5 2.5
Voter Group C: I I C
- - -2.5-- - ---I--- ---3---
3 3.5 4
Voter Group D: I I I
-..-2.5- - - ---2--- ---I---

Key: C = Consistent cognitions, I = Inconsistent cognitions, x = Predicted exposure


rank,- - - x - - - = Potential
information-seeking rank (Rank of 1 = Largest amount).

Some suggestions are possible here for the theoretical integra-


tion of the concept of cognitive consistency with those of infor-
mation-exposure and information-seeking in the election context.
First, it is acknowledged that these data originated from field
surveys rather than experimental evidence and any proposition
of causal connections must be considered tentative. Is it that
the greater exposure has brought about consistent cognitions
between vote intention and election outcome expectations? Or
is it that those with already established consistency subsequently
have chosen to expose themselves to more campaign communica-
tions? The research necessary to resolve this problem first would
identify subgroups of interested voters whose cognitions were
classifiable as consistent or inconsistent early in an election cam-
paign. Repeated measurement of such a panel on exposure to
campaign information as well as subsequent checks on consist-
ency is represented in Figure 1. There, the voter groups have
been designated in an ex post fact0 manner. The amount of pos-
sible information-seeking during each time period is predicted
to be maximal for those whose cognitions attain a consistent
state during that period. For example, Group B in Figure 1 con-
sists of those voters who indicated inconsistent cognitions at T I ,
160 Journal of Communication
but alleviated that inconsistency by the measurement at T,. From
TI to T2, it is expected that their potential information-seeking
was maximal, resulting in a higher exposure level at T2 than
found in those voter groups which remained inconsistent. Across
time, then, the maximum exposure presumably would exist
among those who maintained stable and consistent beliefs, as
found in the present study. Further, the longer the period of
time in which consistent cognitions existed (Group A fits this
description), the greater the predicted exposure to campaign
communication. This paradigm posits that during any single
time period those who hold consistent cognitions are well ex-
posed to campaign information and engage in less seeking, while
those with consistent cognitions and lower exposure levels may
engage in more seeking. The present data, gathered quite close
to the actual election date, may correspond to the situation
depicted from T , to T4.

REFERENCES
1. Berelson, B. P., Lazarsfeld, and W. McPhee. VOTING,University of
Chicago, 1954.
2. Brehm, J. and A. Cohen. EXPLORATION IN COGNITIVE DISSONANCE,
John Wiley and Sons, Inc., New York, 1962.
3. Campbell, A., P. Converse, W. Miller and D. Stokes. THEAMERICAN
VOTER, John Wiley and Sons, Inc., New York, 1960.
4. Carter, R. VOTERSAND THEIRSCHOOLS,Institute for Communication
Research, Stanford University, 1960.
5. Carter, R. “Bandwagon and Sandbagging Effects: Some Measures of
Dissonance Reduction,” PUBLICOPINIONQUARTERLY, 23:279-287.
6. Festinger, L. A THEORY OF COGNITIVE DISSONANCE, Stanford Univer-
sity Press, Stanford, California, 1962.
7. Greenberg, B. “Dimensions of Informal Communication,” in PAUL
J. DEUTSCHMANN MEMORIALPAPERSIN MASS COMMUNICATION RE-
SEARCH ( edited by Wayne Danielson ) Scripps-Howard, Cincinnati,
1963, pp. 35-43.
8. Hyman, H. and P. Sheatsley. “Some Reasons Why Information Cam-
paigns Fail,” PUBLICOPINIONQUARTERLY, 11:412423.
9. Katz, D. (Editor). “Attitude Change,” a special issue of PUBLIC
OPINIONQUARTERLY, 24, Summer, 1960.
10. Klapper, J. THE EFFECTSOF MASS COMMUNICATION, The Free Press,
Glencoe, Illinois, 1961.
11. Maccoby, E., N. Maccoby, A. Romney and J. Adams. “Social Rein-
forcement in Attitude Change,” JOURNAL OF ABNORMALAND SOCIAL
PSYCHOLOGY, 63 ( 1961) :109-115.
12. Maccoby, N. and E. Maccoby. “Homeostatic Theory in Attitude
Change,” PUBLICOPINIONQUARTERLY, 25 ( 1961) 538-545.

You might also like