This paper was presented at the Association for Education in Journalism and
Mass Communication in Toronto, Canada, August 2004.
If you have questions about this paper, please contact the author
directly. If you have questions about the archives, email
[log in to unmask] 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 "").
TO DRILL OR NOT TO DRILL?
ASSESSMENTS OF NEWS COVERAGE AND PUBLIC OPINION
REGARDING U.S. ENVIRONMENTAL POLICIES
Cindy T. Christen, Ph.D.
Department of Journalism and Technical Communication
Colorado State University
C-240 Clark Building
Fort Collins, CO 80523
[log in to unmask]
RUNNING HEAD: To drill or not to drill?
Note. Author citations have been omitted to facilitate blind review. The
author thanks Jessie Ho from Iowa State University for research assistance.
TO DRILL OR NOT TO DRILL?
ASSESSMENTS OF NEWS COVERAGE AND CITIZEN OPINIONS
REGARDING U.S. ENVIRONMENTAL POLICIES
Existing research suggests that both highly involved partisans and those
less involved in environmental issue may judge the climate of opinion
regarding U.S. environmental policies based on their assessments of the
slant of environmental news coverage and their assumption that such
coverage is having a substantial impact on others. To date, however, the
relative effects of locally and nationally circulated news reports have not
been examined. Nor is the potentially countervailing influence of personal
opinion on public opinion estimates well understood. This experiment
examined assessments of the slant and reach of local and national news
articles, as well as the influence of personal opinion, on estimates of
public support for two environmental policy alternatives: oil drilling in
the Arctic National Wildlife Refuge and ratification of the Kyoto treaty on
global warming. While the perceived slant of environmental news articles
proved to be a fairly robust predictor of nonpartisans' public opinion
estimates, support for an effect of perceived media reach on opinion
judgments was mixed. Exposure to news articles that contradicted personal
views appeared to enhance participants' tendency to project those views
onto others. Incorporating level of involvement and issue salience
variables should produce a public opinion inference model that is more
sensitive to the differing influences of perceived news slant, perceived
media reach and personal opinion.
TO DRILL OR NOT TO DRILL?
ASSESSMENTS OF NEWS COVERAGE AND PUBLIC OPINION
REGARDING U.S. ENVIRONMENTAL POLICY
To those who closely follow the news regarding U.S. environmental policy
issues, the headlines are familiar. "U.S. Attacked as EU Ratifies Kyoto"
(CNN.com, June 6, 2001). "Scientists Predict Widespread Extinction by
Global Warming" (The New York Times, January 8, 2004). "Arctic Drilling
Would Set Bad Precedent" (USA Today, October 16, 2003). "Bush: Cut Trees to
Save Forests" (CBSNews.com, May 20, 2003). "Eagle Delisting A Mistake,
Officials Say" (CNN.com, September 27, 1999).
News stories such as these can fuel public interest in decision-making that
impacts the environment and perhaps shape individual preferences regarding
policy alternatives. However, there is another, more subtle effect that
such articles can have on consumers of environmental news, and that is to
shape impressions of what other citizens feel about those same issues and
policy alternatives. To the extent that policy-makers rely on news reports
as an indirect indicator of public sentiment regarding environmental
issues, understanding the ways in which news coverage can influence
impressions of public opinion becomes important to the development of
policies that accurately reflect citizen preferences.
A number of scholars hypothesize a direct effect of news coverage on public
opinion estimates through the inclusion of exemplars and base rate
information in news reports. Studies by Brosius and Bathelt (1994), Gibson
and Zillmann (1994), Zillmann, Gibson, Sundar and Perkins (1996) and
Zillmann, Perkins and Sundar (1992) have shown that vivid first-person
anecdotes tend to be more available in memory than pallid base-rate cues
and thus tend to exert greater influence on opinion judgments. Research
conducted by McLeod and Hertog (1999) is also germane to the question of
opinion estimation and indicates that news stories – by depicting
violations of community law, conformity with social norms and bystander
reactions to protests and strikes – can effectively communicate normative
public opinion to citizens. In addition, opinion-laden news content – such
as guest editorials, "man on the street" interviews and letters to the
editor – can offer insights into public sentiment on environmental and
other topics currently in the news (Gunther, 1998).
A second line of inquiry indicates that news coverage, in addition to
offering direct clues, can influence impressions of public opinion
regarding environmental issues in indirect ways. Gunther and associates
(e.g., Gunther, 1998; Gunther & Christen, 1999, 2002; Gunther, Christen,
Liebhart & Chia, 2001; Christen, Kannaovakun & Gunther, 2002) have
conducted a series of studies seeking to determine if news consumers infer
public opinion in part from their subjective assessments of the slant of
news coverage and their assumptions that numerous others are being exposed
to and influenced by such coverage. According to Gunther (1998), people,
based on exposure to a small and perhaps biased subset of news content,
make judgments about the overall slant of news coverage of particular
issues. Because these individuals perceive news media as having a broad
reach, they also assume that large numbers of others are reading, viewing,
hearing or otherwise being exposed to this same perceived news coverage.
Finally, as predicted by the third-person perception (Davison, 1983),
people tend to exaggerate the influence that such coverage is having on the
opinions of others. Thus, people infer public opinion from the slant of
news stories they believe others are being exposed to.
Past studies of the persuasive press inference have examined a range of
science, environmental and health controversies, including the use of
primates in laboratory experiments, the use of bovine growth hormone in
dairy cattle, genetically modified foods and physician-assisted suicide. In
early experimental tests, neutral participants who read news articles
manipulated to have a favorable or unfavorable slant perceived
corresponding differences in public sentiment on the issues reported.
Experimental data supported the inference hypothesis in conditions both
with and without vivid exemplars (Gunther, 1998), and even when news
stories contained base-rate data that asserted opinions contrary to story
slant (Gunther & Christen, 1999).
Subsequent experimental and survey-based studies demonstrated that highly
involved partisans, like neutral individuals, inferred public opinion in
part from assessments of bias in news reports. Unlike nonpartisans,
however, those who were highly involved in an issue – i.e., those who
closely identified with a particular group and exhibited extreme attitudes
toward other parties (Perloff, 1989) -- were more likely to perceive news
coverage as hostile to their personal views. Evidence from existing studies
of the hostile media effect indicates that people who are highly involved
in an issue are prone to perceiving news coverage as relatively hostile
when content is objectively neutral (Christen et al, 2002; Giner-Sorolla &
Chaiken, 1994; Perloff, 1989; Vallone, Ross & Lepper, 1985) and even when
the news is slanted in favor of or against a particular viewpoint (Gunther
et al., 2001). However, an increased tendency among partisans to
overestimate the number of other Americans who shared their particular
views exerted a countervailing influence on inferences derived from
so-called hostile news, leaving no apparent effect of news assessments on
public opinion judgments (Gunther et al., 2001; Christen et al., 2002).
Thus, among highly involved partisans, projection of personal opinion (also
known as false consensus bias or looking glass effect; see Ross, Green &
House, 1977; Fields & Schuman, 1976) emerges as a second suspect when
considering possible influences on public opinion judgments regarding U.S.
environmental policy issues. According to the projection model, people,
when estimating the distribution of opinions regarding particular
environmental issues, will tend to use their personal views as anchor
points. Selective exposure to similar others, focusing exclusively on one's
own position, seeing oneself and others as similarly rational beings, and
the need to maintain self-esteem and confidence may contribute to
perceptions that other Americans have similar views regarding ANWR oil
drilling, Kyoto treaty ratification and other environmental policy issues
(Marks & Miller, 1987; Christen & Gunther, 2003).
While existing research largely supports the predicted effects of perceived
news slant and personal opinion on public opinion judgments, a number of
important questions remain unaddressed. One concerns the assumption of
media reach; namely, the belief that mass media are conveying similar news
accounts to large numbers of people. While this assumption has strong face
validity, it has yet to be directly tested via a comparison of the effects
of perceived reach across various news sources. One study did examine the
influence of perceived reach on opinion estimates regarding animal
experimentation (see Gunther et al., 2001); however, actual media reach was
a constant, with all news articles formatted to appear as if they had been
clipped from the same national news publications. To fully understand the
influence of perceived media reach, it is important to compare the effects
of perceived news slant under high- and low-reach conditions. If consumers
of environmental news reports do perceive mass media as having a broad
reach, as the persuasive press inference assumes, then the influence of
perceived news slant on estimated public opinion regarding environmental
policy issues should increase with impressions that larger numbers of U.S.
citizens are being exposed to the same news reports.
The influence of involvement on perceptions of news coverage and public
opinion regarding environmental policy issues is also imperfectly
understood. The persuasive press inference model, as currently conceived,
predicts both a direct effect of personal opinion on public opinion
estimates and an indirect effect through subjective assessments of news
slant (see, e.g., Gunther et al., 2001). And while this causal order
appears to hold when considering the opinion judgments of highly involved
partisans (Christen et al., 2002; Gunther et al., 2001), it does not appear
to map well onto the processes by which neutral individuals form
impressions of public opinion.
Rather, early tests of the persuasive press inference hypothesis suggest
that nonpartisans – those who do not closely identify with a particular
partisan group and hold neutral attitudes toward key parties – tend to
perceive the slant of news coverage as congenial with personal views
(Gunther, 1998; Gunther & Christen, 1999). And, we suspect, nonpartisans'
personal views may be a reflection of exposure to news reports, rather than
a persuasive influence on the assessment of such reports. However, existing
research on the persuasive press inference has not tested these competing
conjectures regarding the order of influence directly. Thus, the present
study, in addition to comparing the perceived reach and influence of
locally and nationally circulated news reports on environmental issues,
considers the possibility that the order of influence of perceived news
slant and personal opinion on estimated public opinion regarding U.S.
environmental policies varies based on level of involvement.
Based on past studies of the persuasive press inference, projection and
related phenomena, the following hypotheses are proposed:
H1: Relative Hostile Media Effect – Partisans on opposing sides of
environmental policy issues will perceive the slant of news articles as
relatively more hostile to their particular views.
H2: Persuasive Press Inference
H2a: Among nonpartisans, there will be a positive relationship between the
perceived slant of news articles and estimated public opinion on
environmental policy issues.
H2b: Among partisans, there will be no relationship between the perceived
slant of news articles and estimated public opinion on environmental policy
H3: Perceived Media Reach -- There will be a positive relationship between
the perceived reach of favorable and unfavorable news articles and
estimated public opinion on environmental policy issues.
H4: Level of Involvement
H4a: Among nonpartisans, there will be a positive effect of perceived news
article slant on personal opinion regarding environmental policy issues.
H4b: Among partisans, there will be no effect of perceived news article slant
on personal opinion regarding environmental policy issues.
H5: Projection of Personal Opinion -- There will be a positive relationship
between personal opinion and estimated public opinion on environmental
Data were collected in a 2 x 2 x 2 experimental comparison of local and
national news publications, favorably and unfavorably slanted news
articles, and question order (personal and public opinion questions asked
before or after reading news articles).
As positions regarding environmental policy issues tend to be closely
associated with political party affiliation, Democratic and Republican
Central Committees from the same Midwestern county were approached about
participating in the study. Using mailing lists provided by the respective
committees, a census sample of 75 Republican committee members, and a
random sample of 75 of 106 Democratic committee members, were drawn. An
initial wave of postcards explained the purpose of the study and informed
member that they would be receiving a questionnaire in the mail. The
questionnaires (previously randomized using a random numbers table) were
distributed by mail one week later, with a pre-addressed, postage-paid
return envelope included. To increase response rate, a second wave of
reminder postcards was mailed two weeks following questionnaire
distribution. Thirty-six Democrats and 33 Republicans completed and
returned questionnaires, for a response rate of 46 percent.
In addition, 67 students were recruited from communication courses at a
large Midwestern university. Some students received extra credit for
participation in the study. Questionnaires (previously randomized using a
random numbers table) were administered in person in classroom settings.
Students were subsequently categorized as partisan or nonpartisan based on
responses to attitudinal questions contained in the questionnaire. Forty
students, whose scores ranged from –1.5 to +1.5 on the 11-point attitude
scale, were classified as nonpartisans and their responses considered in
the subsequent analysis.
To camouflage the intent of the experiment, the questionnaire cover sheet
informed participants that researchers were investigating news coverage of
some recent national issues. Participants would be asked to share their
impressions of new reports on one such issue: U.S. environmental policy.
News articles included in the questionnaires addressed two issues: opening
the Arctic National Wildlife Refuge in Alaska to oil exploration, and U.S.
ratification of the Kyoto Treaty to combat global warming.
Prior to reading the two experimental news articles, participants responded
to questions on attitudes towards key parties involved in the issues, and
perceptions of the slant of general news coverage of the two issues. After
reading the two news articles, participants were asked to assess the slant
and reach of the articles. A random half of participants responded to
personal opinion and perceived public opinion questions either prior to or
after reading the two articles, to isolate the effects of perceived news
slant and reach from other influences on opinion estimates. Finally,
demographic data were gathered.
As partisan participants responded by mail, they could not be debriefed
following completion of questionnaires. Instead, a statement regarding the
true nature of the experiment and the source, slant and reach manipulations
was included in the summary of results to be provided to both Democratic
and Republican Central Committees. Student participants were debriefed
about source, slant and question order manipulations immediately upon
completion of questionnaires.
The two purported national news publications were Time Magazine and the New
York Times. The font, layout and general appearance of both news articles
were designed to give the impression that the articles had been clipped
from the actual publications. The front pages of two community college
newspapers in the Midwest were also recreated: the Iowa Lakes College
Spindrift and the Muscatine Community College Calumet. The font, layout
and general appearance were designed to give participants the impression
that they were reading articles that had actually appeared in the two local
To control for the influence of content, the text of each favorably or
unfavorably slanted news article was identical regardless of whether the
story appeared in a locally or nationally circulated news publication.
Articles were constructed from actual news reports; however, headlines and
key words were manipulated to more clearly convey a favorable or
unfavorable slant on the issues. For example, the headline for the
negatively slanted ANWR article read "Bush Under Attack on Drilling," while
the headline for the positive article read "Bush Team Reaffirms Commitment
to Drilling in Arctic Refuge." The headline for the negatively slanted
Kyoto article read "Bush Going Empty-Handed to EU Meeting on Global
Warming," while the headline for the positive article read "Bush Reaffirms
Global Warming Stance Heading to Europe"
To assess level of involvement, measures adapted from Perloff (1989) were
used. Participants were first asked to indicate the political party with
which they identified most closely: Democrat, Republican, Independent,
Other or None. Those who responded Other were asked to provide the name of
the party. Participants' attitudes toward the Democratic and Republican
Parties, George W. Bush and environmental advocacy groups were then
measured using 11-point scales, anchored by –5 (Very Unfavorable Attitude)
and +5 (Very Favorable Attitude), with 0 indicating a Neutral Attitude.
Items assessing attitudes toward the Republican Party and Bush were reverse
coded prior to analysis.
The four attitude measures (including the two recoded items) proved to have
high inter-item consistency (Cronbach's alpha =.86). Responses to the four
attitude questions were also subjected to a principal components factor
analysis followed by varimax rotation and an eigenvalue of 1 as a stop
criterion; a single component including all four attitude items was
extracted. Since discriminant validity was not demonstrated, a single scale
assessing participants' attitudes was constructed by computing the average
score for the four attitude questions, with higher scores indicating
positive attitudes toward the Democratic Party and environmental advocacy
groups and negative attitudes toward the Republican Party and Bush.
To assess perceptions of news article slant, items adapted from Gunther et
al. (2001) were employed. Using 11-point scales, ranging from –5 (Strongly
Biased Against) to +5 (Strongly Biased in Favor), participants evaluated
news article bias toward oil drilling in the Arctic National Wildlife
Refuge and ratification of the Kyoto treaty on global warming. Participants
also shared their perceptions of news article reach by estimating the
percentage of Americans who had read each article. Participants responded
by circling percentage ranges on a 10-point scale, with 1 (0-10%) and 10
(91-100%) as anchor points. Mean scores were later converted to percentages
to facilitate comprehension of results.
Either prior to or after reading the two news articles, participants were
asked to share their own opinions on both environmental policy issues as
well as estimate the views of other Americans. Participants first indicated
(by responding Yes, No or Don't Know) if they personally thought the Arctic
National Wildlife Refuge should be opened to oil drilling. In addition,
participants shared their personal view as to whether the U.S. should
ratify the Kyoto treaty on global warming. After sharing each opinion,
participants indicated how strongly they felt about that opinion by
circling one number on a 10-point scale, ranging from 1 (Not At All Strong)
to 10 (Extremely Strong). Responses to each set of personal opinion and
opinion strength questions were transformed to create a single personal
opinion indicator for each issue, with –10 (Strong No) and +10 (Strong Yes)
as anchor points (Gunther & Christen, 2002).
The set of public opinion items, adapted from Gunther et al. (2001), asked
participants to estimate the percentage of Americans who would respond
affirmatively to those same two opinion questions. Participants estimated
public sentiment by circling percentage ranges on a 10-point scale,
anchored by 1 (0-10%) and 10 (91-100%). Again, mean scores were converted
to percentages to facilitate comprehension of results.
Finally, demographic data were gathered, including age, sex, race,
education, income, political party membership and political ideology. The
latter variable was measured using a 7-point scale, ranging from 1 (Very
Conservative) to 7 (Very Liberal), with 4 indicating Middle of the Road.
Attitude scores confirmed that Democrats held favorable attitudes toward
environmental advocacy groups and the Democratic Party (M=3.04) and
unfavorable attitudes toward Bush and the Republican Party. Similarly, the
attitudes of Republican Party participants were favorable toward Bush and
the Republican Party and unfavorable toward environmental advocacy groups
and the Democratic Party (M=-2.54). Nonpartisan participants differed
significantly from both Republican and Democratic members and held
attitudes slightly favorable toward Bush and the Republican Party (M=-.09).
Across all three groups of participants, differences in attitudes were
highly significant, F(2,105)=115.74, p<.001.
Manipulation checks revealed that news articles favorable or unfavorable to
ANWR oil drilling and Kyoto treaty ratification were rated accordingly by
participants. Mean differences in perceived slant between those who read
news stories favorable (M=1.98) or unfavorable (M= -2.71) toward oil
drilling in the Arctic National Wildlife Refuge were highly significant,
t=11.71, p<.001. Similarly, perceived news article slant among participants
who read stories favoring ratification of the Kyoto treaty on global
warming (M=1.38) differed significantly from those who read stories
opposing ratification (M=-1.04), t=-5.90, p<.001.
First, we examined the responses of Democrats, Republicans and nonpartisan
participants for evidence of a relative hostile media effect. According to
Hypothesis 1, partisans on opposing sides of environmental issues should
see the same environmental news content as relatively more hostile to their
particular positions. Among partisan participants who read news articles
favorable to Bush Administration positions on the two environmental policy
issues (pro-ANWR drilling, anti-Kyoto ratification), independent samples t
tests revealed significant differences in the predicted direction between
Republicans and Democrats. Although both Republican and Democratic Party
members viewed the pro-ANWR article as biased in favor of drilling,
Democrats (M=3.28) judged the article to be significantly more favorable
toward drilling than did Republican participants (M=.27). Differences in
the perceived slant of the pro-drilling article were highly significant,
t=6.16, p<.001. Similarly, both Democrats and Republicans judged the
anti-Kyoto article to be biased against ratification, but members of the
Democratic Party (M=-1.53) judged the article to be significantly more
negative toward treaty ratification than did members of the Republican
Party (M=-.27), t=-2.12, p<.05.
Among partisans who read news articles unfavorable to Bush Administration
positions (anti-drilling, pro-ratification), slant perceptions again varied
in the expected direction. Although both Republican and Democratic
participants concurred that the anti-ANWR article was biased against
drilling, Republicans (M=-3.65) assessed the article as being more negative
than did Democrats (M=-2.67). Similarly, all participants judged the
pro-Kyoto article to be biased in favor of treaty ratification, but members
of the Republican Party (M=2.29) judged the article to be significantly
more favorable than did members of the Democratic Party (M=1.11). For
unfavorable articles, differences in perceived slant between partisans were
marginally significant (ANWR, p=.093; Kyoto, p=.061). Taken together with
other findings, Hypothesis 1 was largely supported.
To test the two persuasive press inference hypotheses, bivariate
correlations were computed, separately for each issue, for those who
responded to public opinion questions after reading news articles on the
issue. Hypothesis 2a proposed a positive association between perceived news
article slant and estimated public opinion on ANWR and Kyoto issues among
nonpartisans. As expected, a significant positive relationship was found
between perceptions of bias in the ANWR news articles and the estimated
percentage of Americans who favored drilling in the refuge, r=.60, p<.05.
However, the positive association between perceived article slant and
estimates that Americans favored ratification of the Kyoto treaty was not
significant. Thus, support for Hypothesis 2a, while indicative of a
positive relationship, was mixed.
Hypothesis 2b predicted that there would be no association between
perceived news article slant and public opinion judgments among partisan
Democratic and Republican participants. As expected, the relationship
between perceptions of bias in the ANWR news articles and the estimated
percentage of Americans who favored drilling in the refuge, while positive,
was not significant, r=-.17, p=.165. The association between perceived
article slant and estimates that Americans favored ratification of the
Kyoto treaty also failed to achieve significance, r=-.13, p=.234. Thus,
Hypothesis 2b was supported.
Hypothesis 3 predicted that the perceived reach of news articles and
estimated public opinion would be positively related. As the perceived
reach of favorably and unfavorably slanted news articles could exert
countervailing influences on opinion estimates, bivariate correlations were
computed within slant conditions for those who responded to public opinion
questions after reading news articles. For the news articles biased against
ANWR drilling, correlational analysis revealed a significant negative
relationship between the perceived reach of the articles and estimates that
Americans opposed drilling: as perceptions of the number of Americans
reading the negative articles increased, estimated support for drilling
decreased, r=-.38, p<.05. However, no relationship was found between the
perceived reach of the favorable ANWR articles and estimates that Americans
supported drilling. On the Kyoto issue, a marginally significant positive
relationship was found between the perceived reach of news articles
favoring treaty ratification and perceived public support for ratification,
r=.35, p=.058. However, the association between the perceived reach of
negatively slanted articles and estimated public support for ratification,
while negative as expected (r=-.20), was not significant. Thus, Hypothesis
3 only received partial support.
Our fourth set of hypotheses tested the conjecture that the order of
influence of perceived news slant and personal opinion would vary based on
level of involvement. To examine this possibility, slant and question order
variables were transformed to create a compound variable with four
conditions: favorable/before (favorable article slant/opinion questions
answered before reading articles), favorable/after, unfavorable/before and
unfavorable after. One-way analyses of variance were then conducted
separately for nonpartisans, Democrats and Republicans.
Hypothesis 4a predicted significant differences in personal opinion among
nonpartisans answering opinion questions before or after reading favorable
or unfavorable news articles. Not surprisingly, differences in personal
opinion among nonpartisans who answered personal opinion questions before
reading ANWR stories were not significant. Among those who responding to
personal opinion questions after reading favorable or unfavorable articles,
however, post-hoc multiple comparisons tests using Tukey's LSD procedure
revealed a significant difference between personal views. While in
aggregate nonpartisans opposed opening the Arctic National Wildlife Refuge
to oil exploration, those nonpartisans who read favorable articles were
more strongly opposed to drilling (M=-7.83) than those who read unfavorable
articles (M=-2.22). Note that personal opinions varied in a direction
opposite that predicted by Hypothesis 4a.
On the issue of Kyoto treaty ratification, personal opinions varied in the
expected direction, with those who responded to opinion questions after
reading favorable articles more likely to support ratification (M=2.00)
than those who read unfavorable articles (M=-.40); however, the differences
observed were not significant. Hypothesis 4a thus received little support.
Among Democrats and Republicans, the effect of perceived news article slant
on personal opinion was expected to be negligible (Hypothesis 4b).
Supporting this prediction, differences in personal views among Democrats
who answered opinion questions before or after reading favorable or
unfavorable news reports on ANWR and Kyoto were not significant. Similarly,
no significant differences in personal opinion were detected between
Republicans who indicated personal views before or after reading articles
unfavorable toward oil drilling in the Arctic refuge. Contrary to
expectations, however, a significant difference in personal opinion was
obtained for Republicans reading news stories that were supportive of ANWR
oil drilling. Those who answered opinion questions before reading the
favorable ANWR article indicated support for oil drilling (M=5.88), while
those who responded after reading the favorable article were slightly
opposed to drilling (M=.33). Again, personal opinions varied in a direction
contrary to expectations. Findings for those Republicans who shared their
personal views after reading the favorable article also differed
significantly from those who answered before (M=8.90) or after (M=8.00)
reading unfavorable ANWR articles. Across all four conditions, significant
differences in personal opinion among Republicans on the ANWR drilling
issue were found, F(3,28)=5.41, p<.01.
While the omnibus F test was not significant for Republican views regarding
Kyoto treaty ratification, post-hoc multiple comparisons tests revealed a
significant difference between those who responded to opinion questions
before reading a negative article on Kyoto ratification (M=-9.00) and those
who answered after reading a positive article on the subject (M=-1.57).
Taken together with other findings, support for Hypothesis 4B was weak and
Hypothesis 5, the projection hypothesis, predicted a positive relationship
between personal opinions and estimated public opinion on environmental
policy issues. To test this hypothesis, bivariate correlations were
computed, separately for each issue, across all participants and separately
for Republicans, Democrats and nonpartisan participants. In aggregate, a
significant positive relationship was found between personal opinion and
estimates that Americans favored drilling in the Arctic refuge, r=.39,
p<.001. Correlational analysis also revealed a moderately strong positive
association between personal opinion and perceived public opinion among
Republicans, r=.45, p<.01. However, the positive relationship between
personal opinion and assessments of public opinion among nonpartisans
narrowly missed the significance threshold, r=.21, p=.096. Contrary to
expectations, a weak negative relationship (r=-.15) was detected between
Democrats' personal views and estimates that the public favored oil
drilling in ANWR; this relationship was not significant, however.
On the Kyoto issue, support for an effect of personal opinion on public
opinion judgments was similarly mixed. Across all participants, the
expected positive relationship between personal opinion and estimates that
Americans favored treaty ratification was detected, r=.17, p<.05. However,
the association between personal opinion and assessments of public opinion
among Republicans and Democrats, respectively – while positive, as expected
– did not achieve significance: Republicans, r=.14, n.s.; Democrats, r=.12,
n.s. No relationship was found between nonpartisans' personal views and
impressions that Americans favored ratification of the Kyoto treaty.
When contemplating possible influences on public opinion estimates
regarding U.S. environmental policy issues, one culprit that quickly
emerges is news reports in mass media. For most Americans, mass media
comprise the primary if not the only source of information on environmental
policy issues. Replicating previous findings (Gunther, 1998; Gunther &
Christen, 1999), the perceived slant of news articles proved to be a fairly
robust predictor of nonpartisans' public opinion estimates. To the extent
that neutral participants perceived environmental news reports as being
favorably or unfavorably biased, their estimates of American sentiment
regarding environmental policy issues tended to vary in the corresponding
direction. Reliable evidence of an effect of perceived news article slant
was obtained for both ANWR oil drilling and Kyoto treaty issues.
Also as predicted, the influence of perceived news article slant on public
opinion judgments by political partisans was negligible. While reliable
evidence of a relative hostile media effect was obtained, perceptions of
hostile bias apparently were not taken into account when Democrats and
Republican assessed support for ANWR oil drilling and Kyoto treaty
ratification. In previous studies, Gunther and associates speculated that
personal opinions might be counteracting the influence of perceived slant
on partisans' public opinion judgments (Gunther et al., 2001; Christen et
al., 2002). When considering findings for order of influence and projection
in the present study, however, the validity of this explanation becomes
Contrary to expectations, partisans' personal opinions did appear to be
influenced by exposure to news articles that supported Arctic oil drilling.
Moreover, they were influenced in a counter-intuitive direction.
Republicans exposed to unfavorable coverage of the ANWR issue were more
extreme in their support for oil drilling than Republicans who read news
reports biased in favor of drilling. Rather than moderating personal views,
exposure to hostile news content apparently led Republicans to adapt more
extreme positions. When news article slant was congenial with existing
views, however, there was no apparent effect of perceived slant on the
personal opinions of Democrat and Republican participants.
A similar trend in personal opinion change was observed for those
participants who were less involved in the issues. While nonpartisans
largely opposed oil drilling in the Arctic refuge, those who read news
articles that advocated oil drilling were more extreme in their opposition
to drilling than nonpartisans who read articles biased against oil
drilling. On the Kyoto issue, however, nonpartisan opinions changed in
expected ways following exposure to favorable or unfavorable news reports.
Taken together, changes in personal opinion among Republicans and
nonpartisans raise two questions: why did exposure to favorable or
unfavorable news reports lead opinions to change in the opposite direction,
and why did the effect of exposure vary between ANWR and Kyoto issues?
With respect to the former question, it appears clear that exposure to news
reports that contradicted personal views triggered a "boomerang" effect,
with both Republicans and nonpartisans becoming more extreme in their
opinions regarding ANWR oil drilling. In a review of research on
projection, Marks and Miller (1987) suggest a possible reason why
nonpartisans may have responded to hostile news articles with greater
adherence to existing positions. As the motivational explanation for
projection suggests, nonpartisans may have been less confident in the
adequacy or correctness of their own position regarding oil drilling in
ANWR. To reduce tension, restore cognitive balance and bolster self-esteem,
nonpartisans may have migrated toward a more extreme stance on the ANWR
issue when confronted with contradictory news reports. For Republican
participants, however, the motivational explanation for opinion
polarization is less satisfactory. Republican Party members, being more
highly involved in the ANWR issue, presumably were more certain of the
appropriateness of their positions and in less need of internal reassurance.
In the case of partisan participants and perhaps nonpartisans as well,
variations in issue salience represent another possible reason for changes
in personal opinion, as well as a possible explanation for the observed
differences in opinion change between ANWR and Kyoto issues. In an
examination of news assessments and perceived public opinion regarding the
outcome of the 2000 U.S. presidential election, Christen, Gunther and
Kiousis (2002) found that high issue salience moderated the effect of
perceived news article slant on opinion estimates, increasing the tendency
of partisans and nonpartisans to rely on personal views when estimating
American views regarding the legitimacy of the Bush victory. Applying
similar logic to the present study, one can argue that the salience of the
two environmental policy issues may have varied, with ANWR oil drilling
remaining relatively more prominent on media and environmental agendas and
Kyoto treaty ratification receiving comparatively less attention in the
news (although this situation has changed somewhat with the approach of the
2004 U.S. presidential election).
Taken together with findings from the Christen et al. (2002) election
study, evidence obtained in the present study suggests that opinion
polarization, the tendency to project and issue salience are all positively
related. As issue salience increases, adherence to existing views also
increases, as does the tendency to rely on personal opinions when
estimating the views held by others. Conversely, the influence of perceived
news slant on public opinion estimates should diminish with increases in
issue salience, particularly among nonpartisans. Thus, personal opinions
regarding ANWR oil drilling – the more salient of the two issues –
exhibited a boomerang effect, while positions regarding Kyoto treaty
ratification varied in the expected direction. This interpretation of study
results is speculative, we acknowledge, and needs to be examined in future
A second focus of the present study was the assumed influence of perceived
media reach. The effect of perceived news article slant on estimated public
support for ANWR oil drilling and Kyoto treaty ratification was expected to
increase with perceptions that larger numbers of Americans were being
exposed to news articles. Evidence of an effect of perceived reach was
obtained for news articles favoring Kyoto treaty ratification and
unfavorable toward ANWR oil drilling; however, the perceived reach of news
articles consistent with current Bush Administration environmental policies
(i.e., opposed to ratification and supportive of drilling) produced
negligible effects. Again, the polarization of personal opinions in
response to news reports sympathetic to Bush Administration positions, and
the enhanced tendency to estimate public opinion using polarized personal
views as anchor points, offer a possible explanation for this one-sided
trend. This may be particularly true among nonpartisans, who tended to
espouse Democratic views (anti-drilling, pro-ratification) and who were
presumably more susceptible to media influence than Democratic or
Nor, based on the study design, can we eliminate the alternate possibility
that differences in the perceived credibility of local and national news
publications accounted in part for the varying influence of perceived news
article reach. Rather than assuming that fewer people were being exposed to
locally circulated news reports, participants may have viewed local
newspapers as less credible sources of environmental news. (See, e.g.,
Kiousis, in press.) To better understand the role of media reach
perceptions in the public opinion inference process, it becomes important
to replicate the present study while controlling for the influence of
source credibility; for example, by utilizing news reports appearing in
print and online editions of the same national news publication. This would
provide a clearer picture of role of perceived reach in the opinion
inference process, while at the same time extending persuasive press
inference research to a new medium (the Internet).
What is clear from the present study is that the persuasive press inference
model, as currently conceived, is oversimplified. The perceived slant of
news articles does appear to shape, change or reinforce personal views
under certain conditions, among nonpartisans and in some instances among
partisans as well. The likelihood that exposure to hostile news reports
will lead to polarization of personal views and an enhanced tendency to
project those views onto others in turn appears to rely in part on the
salience of the issue, with polarization and projection more likely when
issue salience is high and perceived news article slant exerting greater
influence when issue salience is low, particularly among nonpartisans.
Incorporating level of involvement and issue salience variables should
result in a opinion inference model that is more sensitive to the differing
influences of perceived news slant, perceived media reach and personal
opinion on estimated public support for various environmental policy
Brosius, H., & Bathelt, A. (1994). The utility of exemplars in persuasive
communication. Communication Research, 21, 48-78.
Christen, C.T., & Gunther, A.C. (2003). The influence of mass media and
other culprits on the projection of personal opinion. Communication
Research, 30(4), 414-431.
Christen, C.T., Gunther, A.C., & Kiousis, S. (2002). Inferring public
opinion from news coverage: The moderating effect of issue salience. Paper
presented to the Mass Communication Division of the International
Communication Association, Seoul, South Korea.
Christen, C.T., Kannaovakun, P., & Gunther, A.C. (2002). Hostile media
perceptions: Partisan assessments of press and public during the 1997
United Parcel Service strike. Political Communication, 19(4), 423-436.
CNN.com. (2001). U.S. attacked as EU ratifies Kyoto. June 6, 2001.
CNN.com. (1999). Eagle delisting a mistake, officials say. September 27, 1999.
CBSNews.com. (2003). Bush: Cut trees to save forests. May 20, 2003.
Davison, W.P. (1983). The third person effect in communication. Public
Opinion Quarterly, 52, 1-15.
Fields, J. & Schuman, H. (1976). Public beliefs about the beliefs of the
public. Public Opinion Quarterly, 40, 427-448.
Gibson, R., & Zillmann, D. (1994).
Giner-Sorolla, R., & Chaiken, S. (1994). The causes of hostile media
judgments. Journal of Experimental Social Psychology, 30, 165-180.
Gunther, A.C. (1998). The persuasive press inference: Effects of mass media
on perceived public opinion. Communication Research, 25(5), 486-504.
Gunther, A.C., & Christen, C.T. (2002). Projection or persuasive press?
Contrary effects of personal opinion and perceived news coverage on
estimates of public opinion. Journal of Communication, 52(1), 177-195.
Gunther, A.C., & Christen, C.T. (1999). Effects of news slant and base rate
information on perceived public opinion. Journalism & Mass Communication
Quarterly, 76(2), 277-292.
Gunther, A.C., Christen, C.T., Liebhart, J.L., & Chia, S.C. (2001).
Congenial public, contrary press and biased estimates of the climate of
opinion. Public Opinion Quarterly, 65(3), 295-320.
Kiousis, S. (In press). Public trust or mistrust? Perceptions of media
credibility in the information age. Mass Communication & Society.
Marks, G. & Miller, N. (1987). Ten years of research on the false-consensus
effect: An empirical and theoretical review. Psychological Bulletin, 41(1),
McLeod, D.M., & Hertog, J.K. (1992). The manufacture of "public opinion" by
reporters: Informal cues for public perceptions of protest groups.
Discourse and Society, 3, 259-275.
New York Times. (2004). Scientists predict widespread extinction by global
warming. January 8, 2004.
Perloff, R.M. (1989). Ego-involvement and the third-person effect of
televised news coverage. Communication Research, 16(2), 236-262.
Ross, L., Green, D. & House, P. (1977). The "false consensus effect": An
egocentric bias in social perception and attribution processes. Journal of
Experimental Social Psychology, 13, 279-301.
USA Today. (2003). Arctic drilling would set bad precedent. October 16, 2003.
Vallone, R., Ross, L., & Lepper, M. (1985). The hostile media phenomenon:
Biased perception of media bias in coverage of the Beirut massacre. Journal
of Personality and Social Psychology, 49(3), 577-585.
Zillmann, D., Gibson, R., Sundar, S.S., & Perkins, Jr., J.W. (1996).
Effects of exemplification in news reports on the perception of social
issues. Journalism & Mass Communication Quarterly, 73, 427-444.
Zillmann, D., Perkins, Jr., J.W., & Sundar, S.S. (1992).
Note. Author citations have been omitted to facilitate blind review.