Political Communication and ELM
the Mass Media and the
Elaboration Likelihood Model:
Match or Misamatch
2115 Hunter Lane
Columbia, Mo. 65202
Phone: (314) 882-5740; (314) 474-3767
E-mail: [log in to unmask]
[log in to unmask]
Since the dawn of political communication research with Paul F.
Lazarsfeld's Erie County study in 1940 (Lazarsfeld, Berelson & Gaudet,
1944), communication scholars have sought to explain voter behavior and
mass media's role in it. Specifically, researchers have concentrated on
identifying how a voter acquires political knowledge and on
what factors affect a voter's decision to select a candidate or favor
oppose a ballot issue.
Lazarsfeld's studies at Columbia University and subsequent psychological
studies at the University of Michigan found that the mass media had
"limited effects" on voting decisions (Chaffee & Hochheimer, 1985.)
Instead, predetermined factors such as party identification, the party
affiliation of parents and socioeconomic status were more important
predictors of voter behavior (Lowery & De Fleur, 1988).
Since Joseph T. Klapper's watershed book synthesizing the minimal effects
model (Klapper, 1960), researchers have discovered that the mass media
more impact on political communication than was originally believed. In a
1985 article in in the Mass Communication Review Yearbook titled "The
Beginnings of Political Communication Research in the United States,"
Steven H. Chaffee and John L. Hochheimer argue that the empirical support
for the limited effects model was problematic because of the way the
Columbia University studies were designed and conducted (Chaffee &
Hochheimer, 1985). For instance, they point out that the studies were
conducted in a racially homogenous, largely rural and conservative area
Ohio dominated by Republicans. News coverage, they say, also favored
More than two-thirds of the respondents in the initial study mentioned
newspapers and radio as a "helpful" source of political information
less than half mentioned any type of personal source, Chaffee and
Hochheimer point out. Despite these findings, they note that the Columbia
authors concluded that "more than anything else people can move
which became the basis for the two-step flow model of personal
(Katz & Lazarsfeld, 1955).
Although Chaffee and Hochheimer note that a "new flowering" of research
has explored the role of mass communication research in the political
process by breaking out of the Lazarsfeld model in many ways, "most of
assumptions of the original studies continue to structure current
research." They stress the need for new approaches for looking at political
communication without the constraints of prior assumptions.
Two years before Chaffee and Hochheimer formally issued their appeal, Yuko
Miyo took such an approach in a study testing the knowledge gap
hypothesis. The study, conducted at the University of Wisconsin-Madison as
part of the 1980 presidential election campaign and political
project, used a three-way panel design to test how media use would affect
political knowledge (Miyo, 1983). A year earlier, Chaffee and Miyo
presented a paper based on a different aspect of the same project (Chaffee
& Miyo, 1982).
Panel members participated in telephone surveys that were conducted six
months before the election, the month before the election and a year
the election. They were classified according to the type of media that
used and to their amount of media use. Their level of political knowledge
was determined by a combined score of basic political knowledge
and specific questions about the Democratic and the Republican
campaign issues. Their level of education was also determined.
As hypothesized, Miyo found that the level of political knowledge for all
panel members increased over the course of the campaign. She also
expected that the knowledge gap between newspaper readers and
viewers decreased as the election approached and as television
the races and the issues increased. As expected, the gap again widened
year after the election, although all panel members had a higher level
political knowledge than before the election.
In addition, those panel members with the highest media use had the
highest level of political knowledge, a hypothesized result. The level of
political knowledge as expected also increased as the level of
These results showed that media exposure can have an effect on the level
of political knowledge over the course of an election campaign. They
indicated that the level of political knowledge varies with the level
Elaboration Likelihood Model
A model persuasion that could help explain political communication and the
media's role in it has received little attention in mass communication
circles, except for in advertising research (Petty & Cacioppo, 1983;
MacInnis and Jaworski, 1989; Schumann, Petty & Clemons, 1990; James and
Hensel, 1991; MacInnis, Moorman and Jaworski, 1991; Petty, Cacioppo and
Haugtvedt, 1992; Darley and Smith, 1993). Developed in 1981 at the
University of Missouri-Columbia by psychologists Richard E. Petty and John
T. Cacioppo, the Elaboration Likelihood Model essentially uses
and cognitive factors to explain attitude change (Petty & Cacioppo, 1981;
Petty & Cacioppo, 1986).
The Elaboration Likelihood Model, or ELM, assumes that people are
motivated to have "correct" attitudes when presented with a persuasive
message. A correct attitude is one that is correct for the person, i.e.
attitude that fits in with the person's beliefs and other attitudes.
order to form an attitude about a particular message, a person must use
of two routes: the central route or the peripheral route.
To use the central route, the person must have the ability to understand
the message, the motivation to read or view the message, and/or the
opportunity to read or view the message. In a political campaign the
message might be a candidate's position paper or fliers, newspaper or
magazine articles, television or radio broadcasts, candidate speeches and
television or radio commercials.
Using the central route, the person will purposefully evaluate, or
process, the message based on the quality of the arguments that are
presented. During the evaluation, the person will generate thoughts about
the message. If the arguments are strong, the person will generate
thoughts about the message. The greater the number of positive thoughts
that are generated, the greater the attitude change in favor of the
If the arguments are weak, however, the person's attitude generally will
not change. The person might even generate counterarguments against
message, which could result in a more negative attitude, the boomerang
In an era of information overload, ELM assumes that people are "cognitive
misers" who hoard their cognitive resources (Petty & Cacioppo, 1983).
Although motivated to hold correct attitudes, people do not have the time
or cognitive energy to process all of the messages via the central
If a person does not have the ability or motivation to process the
or has not been exposed to it, the peripheral route will be used.
Because the person wants to hold a correct attitude but doesn't want to
exert much energy to form it, the person will survey the setting and
one or more cues on which to base the attitude. The cues can be any
of information, including the candidate's appearance, reputation,
party or what friends say about the candidate. Because a cue is being
used, the person will not evaluate the candidate's arguments. Instead,
person's attitude will be based solely on the quality of the cues.
People will not use central or peripheral processing exclusively. Petty
and Cacioppo envision a continuum of elaboration likelihood. On one
the continuum are people that are highly motivated, have a high
process the message and/or are presented with the opportunity to do
There is a high likelihood that these people will elaborate on the
using the central route. On the opposite end of the continuum are
who have low motivation to process the message, low ability to
it and/or have little exposure to the message. There is a low
that these people will elaborate on the message. Instead, they will
peripheral processing. In the middle is a vast gray area where people
use varying amounts of both types of processing. Little research has
conducted in this area.
Because people using the central route have expended more cognitive energy
in evaluating the message, Petty and Cacioppo say their attitudes will be
more enduring and more predictive of behavior (Petty & Cacioppo,
the other hand, because people using the peripheral route have
little energy in their decision, their attitude change will be
and not very predictive of behavior.
The major question that this study seeks to answer is whether the ELM can
be used to predict the level of political knowledge and cue use of
voters. If the ELM can be shown to make these predictions accurately,
could be a valuable tool in understanding how political knowledge is
and how the mass media fits into the equation.
There have been two scholarly political articles published involving ELM:
one by Petty and Cacioppo and another by Jeffery J. Mondak, a
scientist at Indiana University.
The main purpose of Petty and Cacioppo's study was not to research voting
behavior, but to "refine the contemporary conceptualization of need
cognition, to examine whether the effects of need for cognition on
processing and persuasion observed in previous research are
simply to intelligence, and to test a self-contained individual
assumption in the elaboration likelihood model (Petty & Cacioppo,
Need for cognition is a motivational measure of a person's craving for
cognitive activity, and Petty and Cacioppo have developed a Need for
Cognition Scale to measure it. Their findings have shown that people high
in need for cognition will use central route processing (Petty &
For this study, psychology students in introductory classes at the
University of Iowa completed the scales along with other forms at the
beginning of the semester. The scales were used to obtain subjects with
both high and low need for cognition. In a telephone survey eight weeks
before the 1994 presidential election, subjects were asked questions
political knowledge and voting intentions. After the elections, they
contacted again and asked about voting behavior.
As predicted, Petty and Cacioppo found that subjects high in need for
cognition had more political knowledge than those low in need for
cognition. Also as predicted, they found that the
pre-election attitudes of high need for cognition subjects were better
predictors of voting behavior than the pre-election attitudes of those
subjects low in need for cognition.
In another part of the study, Petty and Cacioppo found that both need for
cognition and intelligence, an ability variable, contributed to higher
message processing. But intelligence accounted for more of the variance.
Overall, Petty and Cacioppo concluded that the attitude-behavior
consistency demonstrated by high need for cognition subjects showed that
cognitive elaboration was at work. As a result, they said the study
supported the ELM.
In the second article, Mondak conducted three experiments to determine the
role of source credibility regarding the public's reaction to U.S. Supreme
Court decisions (Mondak, 1990). In the first experiment, Mondak
manipulated the credibility of the source by varying the use of three
well-known Supreme Court cases as supporting information to back up three
hypothetical cases. Forty-eight students in introductory political science
classes at Indiana University filled out surveys among many forms at
beginning of the semester. Under this low elaboration likelihood
with few arguments available, Mondak found that subjects relied on
credibility as a cue.
In the second experiment, Mondak manipulated source credibility and
argument strength. As predicted, he found that the perceived legitimacy of
a policy increased when arguments were strong and decreased when they
weak. Source credibility was found to be important under low and high
elaboration likelihood conditions. (Petty and Cacioppo argue that under
certain conditions, a variable can serve multiple roles. Here
serves as a cue and an argument.)
In the third experiment, personal relevance, a motivational variable, was
added. For two of the three questions, policy legitimacy increased
high relevance conditions.
Mondak concluded that source credibility can: provide a cue for receivers
who do not process a persuasive message; prompt an ambivalent receiver
examine a persuasive appeal; and be evidence considered during a
assessment of a political message.
Both of these studies tested college students, and neither of them was
conceived primarily as a test of the ELM regarding political
Neither of them mentioned the mass media. To test the ELM's ability to
explain political communication, it was applied in a heated campaign
ballot issue in 1994.
Amendment 7, a statewide Missouri ballot initiative commonly known as
Hancock II, provided a unique opportunity to test the ELM. Formulated
Republican U.S. Rep. Mel Hancock, author of the Hancock I Amendment to
Missouri Constitution, Hancock II would have required voter approval
increases at all levels of government.
The initiative was very complicated, and the ballot language was difficult
to understand. The stakes in the election were very high, causing
University of Missouri system President George Russell to declare a hiring
freeze and to estimate that the jobs of as many as 1,000 university
and staff members could be eliminated if the measure were approved (Meyer,
Legal wrangling contributed to the difficulty in understanding Hancock II.
Four lawsuits were filed concerning the initiative (Godmer, 1994). The
primary lawsuit was one questioning the accuracy of a fiscal note that
approved by the state legislature and appeared on the ballot.
The complexity of the issue made it ideal for a study of the ELM. Unlike a
companion ballot measure authorizing riverboat gambling, some people would
have the ability, motivation and/or opportunity to process messages about
Hancock II and others would not. Therefore, we should be able to test
whether subjects were using central route or peripheral route processing.
Using the ELM, the following four hypotheses were formulated regarding
political communication and Hancock II:
Hypothesis 1: Likely voters who have a low level of elaboration likelihood
will rely on cues while likely voters with a high level of elaboration
likelihood will not.
Hypothesis 2: Likely voters with a low level of elaboration likelihood
will rely on friends, a cue, for information while likely voters who
high level of elaboration likelihood will rely on the mass media, which
will provide information for central processing.
Hypothesis 3: Likely voters with a high level of elaboration likelihood
will use the media more than likely voters with a low level of
Hypothesis 4: Likely voters with a high external motivation will have a
high level of elaboration likelihood.
The Center for Advanced Social Research at the University of
Missouri-Columbia, the Columbia Missourian, KOMU-TV/Channel 8 and
FM conducted a statewide telephone poll from Oct. 22, 1994, to Oct. 26,
1994, of 527 voters who said they were "very likely" to vote in the
general election. These respondents also said they had voted in the
Fifty questions, including some specifically generated for this study,
were formulated based on input from various faculty members at the
of Journalism. The questions were mainly about Hancock II, although
questions about the another ballot measure allowing games of chance on
riverboats were also included.
The responses to three factual questions about Hancock II were used for a
combined measure of political knowledge, the dependent variable.
Miyo's general political knowledge questions, these questions were very
Educational level was based upon a demographic question and was used as a
measure of ability. Motivation was measured by the combined values of
questions about intrinsic motivation. Questions about how much
subjects were paying to newspaper, radio and television news stories
advertisements about Hancock II were used as a measure of opportunity.
types of media were included in this study than in Miyo's study, which
only included television and newspapers.
A hierarchical regression was performed to obtain a measure of elaboration
likelihood. All of the relevant demographic variables except for education
level were entered first. Then education level, the opportunity measure
and the intrinsic motivation measure, all variables associated with
ELM, were entered last. Political knowledge was the dependent variable.
The intercept and the betas significant at the 0.05 level along with the
values of their associated variables were then used to form a equation
elaboration likelihood. An elaboration likelihood score was calculated
each subject. Based on the scores, the subjects were divided into
equal categories of high, medium and low elaboration likelihood.
Chi-square tests were then performed to test association. The first test
included the elaboration likelihood categories and an attention
regarding friends, which were used as cues. The second test used the
elaboration likelihood categories and a question about the main sources of
information about Hancock II. The categories were mass media as the
source of information and friends and other sources. The third test
involved questions about weekly media use. The final test concerned a
measure of external motivation. The categories were employed by a
government agency, a category in which subjects were assumed to be highly
motivated because their jobs were at stake, and other job statuses.
the questions are included in the appendix.
As indicated by Table 1, the regression equation yielded four variables
with significant betas: age, sex, news (the opportunity variable), and
education level. Their betas and values were used to form the following
equation: ELM = -0.362862 + 0.0005790 (age) + 0.171925 (sex) + 0.119137
(edlevel) + 0.058561 (news).
The standardized regression coefficients showed news to be the most
important variable. Its coefficient of 0.28181466 was more than twice as
large as its nearest rival, education level, at 0.1310849. Age and sex
nearly identical coefficients at 0.09610546 and 0.09021624
Hypothesis 1 suggested that likely voters low in elaboration likelihood
would be associated with cue use. As indicated in Table 2, the
test yields a value of 38.944 at a P-value of 0.000.
Hypothesis 2 suggested that voters low in elaboration likelihood would be
associated with cue use and those voters high in elaboration
would be associated with mass media use. The chi-square test gives a
of 14.334 and a P-value of 0.001.
Hypothesis 3 suggested that elaboration likelihood would be
associated with the frequency of media use. The more frequent a subject
used the media, the more that subject would be associated with
central-route processing. The chi-square test for newspaper use had a
of 34.750 and a P-value of 0.002 while the test for network television
news use had a value of 41.056 and a P-value of 0.000. The chi-square
for local television news use yielded a value of 30.039 and a P-value
Hypothesis 4 suggested high external motivation would be associated with
high elaboration likelihood. The chi-square test had a value of 24.342
a P-value of 0.007.
Interestingly, a chi-square test of elaboration likelihood and a question
about the perceived difficulty of understanding Hancock II yielded a
of 16.523 and a P-Value of 0.011, what the ELM would predict.
The results of the regression analysis show that opportunity, i.e.
attention to the mass media, and education level were important factors in
determining the level of political knowledge. All of the demographic
factors combined except education level explained only 0.0196 percent of
the variance in political knowledge, as Table 2 indicates. The
education level nearly doubled than amount to 0.0361 percent, a
that shows the impact of ability.
But putting opportunity into the equation boosted the total to 0.1177, an
increase of more than 8 percent of the explained variance. This
evidence that the mass media can play an important role in the
political knowledge. Under high elaboration conditions, this knowledge
will be used to process the message and apparently to help make voting
The addition of the intrinsic motivation variable only accounted for less
than 1 percent of the explained variance and did not have a
beta. In this case, intrinsic motivation alone might not have been
sufficient to induce subjects to process the difficult message via the
The results of the chi-square tests add more support for the role of
media. The tests showed that mass media are associated with high levels
elaboration likelihood, and high levels of media use are associated
high levels of elaboration likelihood. Both results are consistent
Conversely, the chi-square tests indicated that cue use is associated with
low elaboration likelihood and that the use of friends and other nonmedia
sources of information are also associated with low elaboration
Again, these findings are consistent with the ELM.
Finally, a chi-square test showed that high extrinsic motivation is
associated with high elaboration likelihood, another finding that is
consistent with the ELM.
This study has applied the ELM to an area in which it had not been applied
before, i.e. mass media use and political communication. It has also
applied a different research technique, multiple regression, to develop a
measure to identify central and peripheral processors. And it then
resulting model to test the use of ELM in several areas.
In future studies, the use of this model should be expanded. If a survey
technique is used, the questions should be designed to test the ELM
exclusively and more questions about other possible cues and information
sources should be included. The studies could also have a panel design
similar to Miyo's study in order to build on Petty and Cacioppo's
into the predictive power of the ELM regarding actual voting. Such
should also track changes in central and peripheral processing during
after the campaign.
Follow-up interviews should also be conducted with respondents in an
attempt to pinpoint what information sources and cues had the most
on their decisions in the polling booth.
Chaffee, Steven H., and Hochheimer, John L. (1985). "The
Beginning of Political Communication Research in the United
States. Mass Communication Review Yearbook, 75-104.
Chaffee, Steven H. and Miyo, Yuko (1982). "Selective Exposure and
the Reinforcement Hypothesis in the 1980 Presidential
Campaign: An Intergenerational Study." Paper presented to
the Association for Education in Journalism, Athens, OH.
Darley, W.K., Smith, R.E. (October 1993). "Advertising Claim
Objectivity _ Antecedents and Effects." Journal of
Marketing. 57, 100-113.
Godmer, Adam (1994, October 20). Amendment 7 Creates Twisted
Legal Labyrinth. Columbia Missourian, p. 1.
James, Karen E., and Hensel, Paul J. (June 1991). Negative
Advertising: The Malicious Strain of Comparative
Advertising." Journal of Advertising, 20, 53-69.
Katz, Elihu, and Lazarsfeld, Paul F. (1955). Personal Influence.
New York: Free Press.
Klapper, Joseph T. (1960). The Effects of Mass Communication. New
York: Free Press.
Lazarfeld, Paul F., Berelson, Bernard, and Gaudet, Hazel
(1944). The People's Choice: How the Voter Makes Up His Mind
in an Election Campaign. New York: Duell, Sloan and Pearce.
Lowery, Shearon A. and DeFleur, Melvin L. (1988). Milestones in
Mass Communication Research. New York: Longman.
MacInnis, D.J., and Jaworski, B.J. (1989, October). "Information
Processing from Advertisements: Toward an Integrative
Framework." Journal of Marketing, 53, 1-23.
MacInnis, D.J., Moorman, C., and Jaworski, B.J. (1991) "Enhancing
and Measuring Consumers' Motivation, Opportunity, and
Ability to Process Brand Information from Ads." Journal of
Marketing, 55, 32-53.
Meyer, Amy M. (1994, October 4). Russell Predicts U.M. Layoffs.
Columbia Missourian, p. 1.
Miyo, Yuko (1983). "The Knowledge Gap Hypothesis and Media
Dependency." Communication Yearbook #7. 625-650.
Mondak, Jeffery J. "Perceived Legitimacy of Supreme Court
Decisions: Three Functions of Source Credibility."
Political Behavior. 12: 363-384.
Petty, Richard E., and Cacioppo, John T. (1981). Attitudes and
Persuasion: Classic and Contemporary Approaches.
Dubuque, IA: Brown.
Petty, Richard E., and Cacioppo, John T. (1983). "Central and
Peripheral Routes to Advertising Effectiveness: The
Moderating Role of Involvement." Journal of Consumer
Research. 10, 135-146.
Petty, Richard E., and Cacioppo, John T. (1986) Communication and
Persuasion: Central and Peripheral Routes to Attitude
Change. New York: Springer.
Petty, Richard E., Cacioppo, John T., Kao, Chuan Feng, and Rodriguez,
Regina (November 1986. "Central Peripheral Routes
to Persuasion: An Individual Difference Perspective."
Journal of Personality and Social Psychology. 51 132-1043.
Petty, Richard E., Schumann, David W., and Clemons, D. Scott
(1990). "Predicting the Effectiveness of Different
Strategies of Advertising Variation: A Test of the
Repetition-Variation Hypothesis." Journal of Consumer
Research. 17, 192-202.
Petty, Richard E., Cacioppo, John T. and Haugtvedt, Curtis P.
(1992). "Need for Cognition and Advertising: Understanding
the Role of Personality Variables in Consumer Behavior."
Journal of Consumer Psychology. 1, 239-260.
Political Communication and ELM
Appendix POLL QUESTIONS
Hancock Two would require voter approval of any tax
Hancock Two would prohibit large state tax increases from
one year to the next whether approved by the voters or
Hancock Two sets a limit on how high state revenue can go up
each year without a vote of the people.
Introduction: Different people like the news for different
reasons. I'm going to read a list of some reasons for paying
attention to the news. As I do, please tell me whether you
strongly agree, agree, disagree, strongly disagree or have
no opinion about each reason.
To keep up with the latest news.
To figure out what is important.
Cue (Attention Measure)
On a scale of 1 to 5 where 1 in not at all and 5 is a lot,
how much attention are you paying to what your family,
friends and colleagues are saying about Hancock Two?
On a scale of 1 to 5 where 1 is not at all and 5 is a lot
how much attention are you paying to newspaper coverage of
On a scale of 1 to 5 where 1 is not at all and 5 is a lot,
how much attention are you paying to TV news coverage of
On a scale of 1 to 5 where 1 is not at all and 5 is a lot,
how much attention are you paying to radio news and talk
shows about Hancock Two?
On a scale of 1 to 5 where 1 is not at all and 5 is a lot,
how much attention are you paying to television commercials
about Hancock Two?
Mass Media/ Not Mass Media
If you had to choose ONE source to help you understand
complex local or state issues, would you pick . . .
Word-of-mouth, Or some other source? (not mass media);
television commercials, television news, radio, newspapers
Use of Mass Media
On average, how many days in a week do your read a
On average, how many days in a week do you watch the
national network news on TV?
On average, how many days in a week do you watch the local
news on TV?
Extrinsic Motivation (potential lost job)
Is the primary wage earner in your family ...
Employed by a government agency, self-employed, employed by a private
company, employed by a government agency, unemployed, retired or