Communication Research

http://crx.sagepub.com/

A Spiral of Silence for Some: Attitude Certainty and the Expression of

Political Minority Opinions
Jörg Matthes, Kimberly Rios Morrison and Christian Schemer
Communication Research 2010 37: 774 originally published online 16 June 2010
DOI: 10.1177/0093650210362685

The online version of this article can be found at:

http://crx.sagepub.com/content/37/6/774

Published by:

http://www.sagepublications.com

Additional services and information for Communication Research can be found at:

Email Alerts: http://crx.sagepub.com/cgi/alerts

Subscriptions: http://crx.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Citations: http://crx.sagepub.com/content/37/6/774.refs.html

>> Version of Record - Dec 1, 2010

OnlineFirst Version of Record - Jun 16, 2010

What is This?

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

C

Communication Research
37(6) 774­–800
A Spiral of Silence for © The Author(s) 2010
Reprints and permission: http://www.
Some: Attitude Certainty sagepub.com/journalsPermissions.nav
DOI: 10.1177/0093650210362685
and the Expression of http://crx.sagepub.com

Political Minority Opinions

Jörg Matthes,1 Kimberly Rios Morrison,2

and Christian Schemer1

Abstract
Spiral of silence theory does not assume a simple relationship between opinion climate
and opinion expression. In fact, the notion of hardcore individuals (who express their
opinions regardless of the climate) embraces the idea that there are some people for
whom this relationship does not hold true. However, this idea has not been put to a direct
empirical test. In this article, the authors propose that attitude certainty is a key variable
in identifying the hardcore. Data from three surveys demonstrate that the climate of
opinion only determines opinion expression when individuals hold their attitudes with
low or moderate attitude certainty. For individuals with high attitude certainty, no such
effect can be found.Thus, there is a spiral of silence only for some but not for all members
of the public.

Keywords
spiral of silence, opinion climate, opinion expression, attitude certainty

The freedom to express one’s political opinion is one of the fundamental principles of
modern democracy. Ideally, political opinion expression is a process that excludes no one
in a society, no matter how prominent or marginal a particular issue position is. The ques-
tion of who chooses to speak out and who remains silent is thus of fundamental relevance
to mass communication and public opinion research. In answering this question, Noelle-
Neumann’s (1974) spiral of silence theory has played a very prominent and influential role.

1
University of Zurich, Zurich, Switzerland
2
Ohio State University, Columbus

Corresponding Author:
Jörg Matthes, University of Zurich, Institute of Mass Communication and Media Research,
Andreasstrasse 15, 8050 Zurich, Switzerland
Email: [email protected]

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

Matthes et al. 775

According to spiral of silence theory, individuals who perceive their opinions to be in a

majority position are more likely to express their views than those who feel that their opin-
ions are not shared by most other people. The proposed reason for this process is a fear of
social isolation, which leads the former group to speak out and the latter group to fall silent
in spite of their actual distribution in a society.
Although a meta-analysis of spiral of silence studies reveals a significant relationship
between perceived majority status of one’s opinion (i.e., opinion climate) and willingness to
express that opinion, this relationship is rather small in magnitude (Glynn, Hayes, & Shana-
han, 1997; for a discussion, see Scheufele & Moy, 2001). There are a number of possible
reasons for that, such as the failure to control for fear of social isolation (Ho & McLeod,
2008; Neuwirth, Frederick, & Mayo, 2007), the neglect of other individual difference vari-
ables that might be at work (e.g., Hayes, Glynn, & Shanahan, 2005; Miller & Morrison, 2009;
Willnat, Lee, & Detenber, 2002), or the lack of consonance in media coverage (Peter, 2004).
Another explanation for why the relationship between opinion climate and opinion
expression is rather weak has been addressed by Noelle-Neumann (1974) herself: Some
people (the hardcore) choose to speak out regardless of the climate of opinion. The hard-
core appear to be those individuals who hold strong opinions. As a consequence, they are
not afraid of expressing minority views. As several scholars have argued, attitude
strength—in particular, attitude certainty—is a key variable in identifying the hardcore
(e.g., Glynn & McLeod, 1984; Lasorsa, 1991; Neuwirth et al., 2007, Oshagan, 1996;
Salmon & Oshagan, 1990). More specifically, those who are certain of their opinions will
always speak out no matter how hostile the climate of opinion is. In contrast, those who are
rather uncertain fall victim to the spiral of silence. Surprisingly, this idea has not been
directly tested in previous research. Although we know that attitude certainty and related
constructs predict opinion expression (e.g., Baldassare & Katz, 1996; Lasorsa, 1991), we
do not know whether and how the relationship between opinion climate and willingness to
speak out depends on attitude certainty.
The aim of this article is to introduce attitude certainty as a moderator of the spiral of
silence. First, we review previous studies that have included attitude certainty and related
measures to predict the willingness to speak out. After that, we discuss the role of attitude
certainty in spiral of silence theory. We then present the results of three independent studies
that support the claim that attitude certainty moderates the relationship between the climate
of opinion and opinion expression.

Explaining Willingness to Speak Out

The willingness to express one’s opinion is the primary dependent variable in spiral of
silence research. In explaining opinion expression, the perceived climate of opinion serves
as the major independent variable. However, the climate of opinion may not be the only
variable that impacts opinion expression. For instance, a number of scholars have argued
that attitude strength, defined as the extent to which an attitude is resistant to change and
predictive of behavior (Krosnick & Petty, 1995), influences the readiness to speak out.

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

776 Communication Research 37(6)

Specific dimensions of attitude strength include but are not limited to extremity, intensity,
personal interest or importance, and certainty (Krosnick, Boninger, Chuang, Berent, &
Carnot, 1993).
In a study about the issue of abortion, Salmon and Neuwirth (1990) asked Wisconsin
residents “how concerned they were about the issue” and “how strongly they held their
opinion” (p. 572). An index of these two items significantly predicted both willingness to
give an interview to a TV reporter and likelihood of entering a discussion with a stranger
who disagrees. In a similar vein, Moy, Domke, and Stamm (2001) found that perceived
issue importance is a significant predictor of the willingness to speak out on affirmative
action. Likewise, Kim, Han, Shanahan, and Berdayes (2004) found that personal interest
in the issue of Korean unification was the strongest predictor of the willingness to express
one’s views about the unification in a hypothetical TV interview.
In addition to issue interest and personal importance, one’s willingness to express an
opinion can be influenced by the extremity or self-reported intensity of that opinion. For
example, Baldassare and Katz (1996) found that people with extreme political orientations
were more willing to speak out than people with more moderate political orientations.
Using similar measures, Shamir (1997) showed that attitude extremity was significantly
related to the willingness to publicly express an opinion about Israeli territories. Yet
another measure of attitude strength (specifically, intensity) was applied by Neuwirth et al.
(2007), who asked respondents how strongly they felt about the U.S. invasion of Iraq. This
measure was found to be a significant predictor of opinion expression.
Taken together, the studies reviewed above have embraced the notion that opinion
expression does not only depend on opinion climate but also on how strongly one feels
about an issue. Although the results of these studies are fairly consistent, they all treat
attitude strength as a direct predictor of opinion expression, regardless of opinion climate.
What can be derived from spiral of silence theory, however, is that attitude strength—in
particular, attitude certainty—moderates the relationship between opinion climate and
opinion expression. That is, spiral of silence theory actually posits that for the hardcore
(i.e., people who are certain of their attitudes), opinion climate should have no impact on
opinion expression. In contrast, for the rest of the public (i.e., people with low or moderate
attitude certainty), opinion climate should be a crucial determinant of opinion expression.
Thus, examining the moderating role of attitude certainty can shed light on when opinion
climate will and will not affect expression and as such on why the direct relationship
between climate and expression is relatively small (see Glynn et al., 1997).

Attitude Certainty
The construct of attitude certainty, as an indicator of attitude strength, is based on the idea
that people hold their attitudes with varying levels of conviction (Tormala & Rucker,
2007). It is well established that attitude certainty is conceptually and empirically distinct
from other dimensions of attitude strength, such as importance (Krosnick et al., 1993) or
extremity (Prislin, 1996). Although these dimensions are correlated, they load onto sepa-
rate factors (see Krosnick et al., 1993).

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

Matthes et al. 777

The Origins of Attitude Certainty

Numerous studies have demonstrated that attitude certainty is increased by factors, such as
accessibility of the attitude (e.g., Bassili, 1996), direct experience with the attitude object
(e.g., Krosnick et al., 1993), subjective ease of recalling attitude-relevant information (e.g.,
Haddock, Rothman, Reber, & Schwarz, 1999), online attitude formation (e.g., Bizer, Tormala,
Rucker, & Petty, 2006), subjective impressions of successfully resisting compelling per-
suasive messages (e.g., Tormala & Petty, 2002), or the amount, cognitive elaboration, and
structural consistency of attitude-relevant knowledge (e.g., Smith, Fabrigar, MacDougall, &
Wiesenthal, 2008). Other research suggests that intensive elaboration of attitude-relevant
information—which is associated with high attitude certainty—is more likely when indi-
viduals are able and motivated to process the information (Eagly & Chaiken, 1993; Mackie
& Asuncion, 1990). Still other studies suggest that individuals high in the trait need to
evaluate are more likely to hold their attitudes with certainty (Britt, Millard, Sundare-
swaran, & Moore, 2009; Matthes, Wirth, & Schemer, 2007; Tormala & Petty, 2001).
It is important to note that attitude certainty is bound to specific attitude objects. One
can be certain about one object, yet feel uncertain about another. Although there is trait-
based variance—for instance, in the need to evaluate—attitude certainty need not be stable
over time. It can decrease or increase depending on the antecedents discussed above (see
Tormala & Rucker, 2007).

Attitude Certainty and Hardcore Individuals

According to previous spiral of silence research (e.g., Lasorsa, 1991; McDonald, Glynn,
Kim, & Otsman, 2001), hardcore individuals can be identified primarily by their convic-
tion that the attitudes they hold are correct, which is a key component of attitude certainty
(Krosnick et al., 1993; Petrocelli, Tormala, & Rucker, 2007). We argue that—compared to
other attitude strength measures—attitude certainty may be a more accurate way of capturing
the difference between those individuals who constitute the hardcore and those who do not.
Taking attitude extremity, for example, one could hold a moderate attitude toward
health care reforms but feel certain that this is the right way to think about the issue
(Tormala & Rucker, 2007). Similarly, one could be highly interested in a recent issue (e.g.,
the world financial market crisis of 2008 and 2009) but not yet be certain of what opinion
is the correct one to have. According to spiral of silence theory, it is not necessarily being
interested in an issue or having an extreme attitude that makes hardcore people voice
minority opinions. Instead, it is the conviction that their opinion on the issue is correct. In
the one study to date that has examined the relationship between attitude certainty and
opinion expression, Lasorsa (1991) found that attitude certainty did indeed predict respon-
dents’ willingness to express their opinions on several different issues (e.g., AIDS, the
economy), though the moderating role of opinion climate was not tested.
Like in Lasorsa’s (1991) study, most studies have looked at the main effect of attitude
certainty (and related measures) on opinion expression and not at its interaction with opinion
climate. To our knowledge, there are only two exceptions to this. In an experimental

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

778 Communication Research 37(6)

investigation, Oshagan (1996) found that attitude extremity predicted a greater willingness
to express one’s opinion. Even more important for the purposes of the present article, a
significant interaction effect showed that the attitude of one’s friends (as an indicator of
opinion climate) was only a significant predictor of opinion expression for individuals with
low or moderate attitude extremity but not for individuals who reported extreme
attitudes.
Further evidence comes from social psychological research on conformity in group
situations. Hornsey, Majkut, Terry, and McKimmie (2003) found that individuals who did
not have a moral basis for their attitudes were more likely to express their views when they
were in a majority than when they were in a minority position. However, the opposite was
true for individuals who did have a moral basis for their attitudes. Such individuals were
even more willing to publicly voice their attitudes when they were in a minority position
compared to a majority position.

Hypotheses
Taken together, there is some preliminary evidence from two sets of experiments (Hornsey
et al., 2003; Oshagan, 1996) that the relationship between majority perception and opinion
expression may differ across individuals. However, none of these experiments directly
measured attitude certainty. They instead assessed closely related constructs, such as atti-
tude extremity. Furthermore, experimental studies have clear advantages when it comes to
internal validity. However, surveys—traditionally the method of choice in spiral of silence
research—can add external validity and are thus at least equally important.
In line with the notion of the hardcore in spiral of silence theory, we assume that attitude
certainty—over and beyond other strength measures—moderates the relationship between
opinion climate and opinion expression. This leads to our main hypothesis:

Hypothesis 1: For people with uncertain attitudes—but not for people with certain
attitudes—a hostile opinion climate will lead to a decrease in opinion expression.

In addition to this hypothesized interaction, we nevertheless expect a main effect of

attitude certainty on the willingness to speak out. In fact, previous research has repeatedly
demonstrated that individuals who are generally certain in their opinions are more likely to
express their opinions than those who are less certain (e.g., Baldassare & Katz, 1996;
Lasorsa, 1991).

Hypothesis 2: People with certain attitudes are generally more likely to express their
views than people with uncertain attitudes.

Overview of Studies
We present data from three surveys that provide systematic evidence for the role of attitude
certainty in the spiral of silence. In all three studies, we expected to find a significant

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

Matthes et al. 779

interaction between attitude certainty and opinion climate when explaining opinion expres-
sion. In Study 1, opinion climate was operationalized as the congruency between respon-
dents’ own opinion and perceived public opinion. Study 2 replicated the findings of Study
1 with a direct and thus more precise measure of opinion climate. Furthermore, Study 2
explored the role of importance as an alternative moderator. Study 3 was designed to rep-
licate the findings of Studies 1 and 2 using a multi-item measure of opinion climate and a
measure of dangerous public opinion expression as a theoretically more precise test of
spiral of silence theory. All three studies dealt with issues that have a moral component, a
prerequisite for spiral of silence research (Noelle-Neumann, 1993).
To avoid underspecified regression models (for this argument, see Hayes, Scheufele, &
Huge, 2006), and to make the studies comparable, we included the same list of controls in
all three studies. By doing so, we followed Salmon and Neuwirth (1990) who stressed the
importance of including multiple controls “to determine whether the relationship between
opinion climates and expression is spurious or [ . . . ] actually stronger than has been dem-
onstrated in previous bivariate research” (p. 571).
Besides demographics, we controlled for TV and newspaper use, respondents’ own
issue attitudes, and their general frequency of discussion. Media use, as a central variable
in spiral of silence research, has been found to be a direct predictor of opinion expression
in numerous studies (see Baldassare & Katz, 1996; Moy et al., 2001; Neuwirth, 2000;
Neuwirth et al., 2007; Pan, Shen, Paek, & Sun, 2006; Scheufele, 1999). As Baldassare and
Katz (1996) argue, people who follow the news are more confident in their political deci-
sions and are better able to recall their opinions compared to people who do not follow the
news (see also Moy et al., 2001). Likewise, Scheufele (1999) and Noelle-Neumann (1993)
point to the articulation function of mass media: “The media provide people with the words
and phrases they can use to defend a point of view” (Noelle-Neumann, 1993, p. 173). Thus,
it can be expected that media use has a positive impact on opinion expression.
Similarly, controlling for a person’s own opinion appeared warranted because the effect
of opinion climate on opinion expression should be independent from one’s actual position
on an issue (see e.g., Baldassare & Katz, 1996; Shamir, 1997). It is important to ensure that
it is not the actual opinion of respondents but their perceived minority or majority status
that leads them to speak out or to fall silent (see also Neuwirth et al., 2007; Neuwirth &
Frederick, 2004).
Finally, general interpersonal discussion frequency was included in all three studies.
The key dependent variable in spiral of silence theory is opinion expression under situa-
tions of social pressure. This variable, however, must be carefully separated from political
talk or general discussion (Scheufele, 1999). Spiral of silence theory, in fact, does not
assume an effect of opinion climate on general discussion. Thus, it is important to ensure
that the effect of opinion climate on opinion expression is independent from people’s gen-
eral tendency to engage in discussions (Scheufele, 1999). It can be expected that general
discussion frequency has a positive impact on opinion expression, as people who talk more
frequently about politics are also more likely to explicitly express their views (for evi-
dence, Neuwirth, 2000; Pan et al., 2006; Scheufele, 1999).

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

780 Communication Research 37(6)

Study 1
Method
In Study 1, data were gathered by a standardized Web-based survey conducted in January
and February 2005. The survey dealt with attitudes toward the bilateral relations between
Switzerland and the EU. In particular, respondents answered questions about the pros and
cons of Switzerland joining the EU. At the time, this was a major and controversial topic
in Switzerland leading to a political and public debate. Switzerland was (and still is) not a
member of the EU. Under the Swiss system of direct democracy, joining the EU would
entail a national referendum with a majority of Swiss residents approving the membership
application.
Participants in the study (N = 688, 47.8% female, age range: 18-81 years, M = 33.69,
SD = 13.85, nationality: German-speaking Swiss residents) were contacted by e-mail with
the help of mailing lists. Care was taken to produce variability in education and age. Nev-
ertheless, younger people (70% were under 40 years of age) and people with higher educa-
tional degrees were overrepresented (26% college degree, 40% high school degree that is
necessary to go to a college).

Measures
The dependent variable, willingness to express one’s opinion, was measured by several
questions. Regarding the question of Switzerland joining the EU, respondents were asked
whether (a) they express their opinion among friends, family, or colleagues; (b) they par-
ticipate in discussions in public; (c) they participate in demonstrations; and (d) they take
part in collections of signatures. These four items were summed up to an index ranging
from 0 to 4 (M = 1.40, SD = 0.73). Attitude certainty was assessed with a standard item
asking, “How certain are you in your opinion about Switzerland joining the European
Union?” (M = 3.70, SD = 1.01), ranging from 1 (very uncertain) to 5 (very certain). This
item has been successfully used in psychological research and is generally regarded as a
reliable measure of the construct (to name a few, Fazio & Zanna, 1978; Krosnick &
Schumann, 1988; Petrocelli et al., 2007; Tormala & Petty, 2002).
As is common in many spiral of silence studies (e.g., Neuwirth, 2000; Scheufele,
Shanahan, & Lee, 2001; Willnat et al., 2001), majority opinion climate was operational-
ized by combining two measures: First, respondents were asked how they would vote if
there was a referendum about joining the EU the next day (1 = oppose to 5 = in favor; M =
3.07, SD = 1.68). Second, they were asked how likely they thought it was that Switzerland
would be joining the EU—in other words, that the Swiss people would vote in favor of
joining in a referendum (1 = very unlikely to 5 = very likely; M = 2.66, SD = 1.17). These
two items were used to compute a majority/minority index: If respondents opposed joining
the EU and felt that the public would decide the same way or if they were in favor of join-
ing and thought this was going to happen, they were given the code majority climate (2 =
majority, n = 284). In contrast, when they opposed but saw the public in favor or when
they were in favor but thought the public would oppose, they were given a minority position code

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

Matthes et al. 781

(1 = minority, n = 212). Respondents that chose the middle category were excluded because
they could not be assigned to a majority or minority group.
As controls, we measured sex (2 = female, 1 = male), age, and education. Frequency of
TV use (M = 3.43, SD = 1.33) and newspaper use (M = 4.34, SD = 1.00) in regard to poli-
tics were measured on a 5-point scale. Issue attitude was assessed by asking how people
would vote if there was a referendum next Sunday about joining the EU or not (1 = in favor
to 5 = against; M = 3.07, SD = 1.68). General discussion frequency was gauged with the
item, “How often do you talk with other people about Switzerland joining the European
Union?” (1 = very seldom to 5 = very often; M = 2.72, SD = 1.03).
Finally, to explore the role of other (noncertainty) attitude strength measures, we opera-
tionalized attitude extremity as the absolute value of the difference between participants’
own attitude and the scale’s midpoint (see Lavine, Borgida, & Sullivan, 2000). Scores
could range from 0 to 2, with higher scores reflecting more extreme attitudes (M = 1.26,
SD = 0.71).

Data Analysis
Data were analyzed by a blockwise hierarchical multiple regression. Sex, age, and educa-
tion were included in the first block; TV and newspaper use in the second; issue attitude
and general discussion in the third; attitude certainty and majority climate in the fourth;
and finally, the interaction term in the fifth block. To compare our two attitude strength
measures, we ran the identical model with attitude extremity instead of attitude certainty.
Because both concepts were highly correlated in this study (r = .49, p < .001), they could
not be included in one model. Multicollinearity would lead to an overfitting of the regres-
sion model.
In order to probe and interpret the interaction, we used the modprobe macro to apply the
Johnson-Neyman technique (Bauer & Curran, 2005; Hayes & Matthes, 2009; Preacher,
Rucker, & Hayes, 2007). The Johnson-Neyman technique identifies regions in the range of
the moderator variable where the effect of the focal predictor on the outcome is statistically
significant and not significant. This way, it can be precisely determined at which levels of
attitude certainty there is a positive, negative, or no relationship between climate and opin-
ion expression. The Johnson-Neyman technique is superior to the more common pick-a-
point approach that merely picks representative values (e.g., high, moderate, and low) of
the moderator variable (Bauer & Curran, 2005). A difficulty of the pick-a-point approach is
that there are no nonarbitrary guidelines for picking the points at which to probe the interac-
tion (Hayes & Matthes, 2009). Independent variables were not mean centered (see Hayes,
2005). When an interaction was present, the (conditional) main effects of the predictors
were not interpreted. These effects were probed with the Johnson-Neyman technique.

Results
The results of the blockwise multiple regression are depicted in Table 1. As can be seen,
age (β = .11, p < .05), TV (β = .20, p < .001), newspaper use (β = .12, p < .05), and overall

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

782 Communication Research 37(6)

Table 1. Regression Coefficients for Predicting Opinion Expression (Study 1)

Unstandardized Standardized
Predictors Coefficients Coefficients p Value
Block 1: Demographics
  Age .01 .11 **
  Sex −.05 −.03
  Education −.01 −.01
  Incremental R2 = .01
Block 2: Media use
  Frequency TV news .11 .20 ****
  Frequency newspaper use .10 .12 **
  Incremental R2 = .06***
Block 3: Issue attitude and discussion
  Issue attitude .01 .01
  Overall discussion frequency .20 .28 ****
  Incremental R2 = .07***
Block 4: Majority and certainty
  Perceived majority climate .03 .02
  Attitude certainty .05 .08
  Incremental R2 = .00
Block 5: Focal predictors
  Perceived Majority Climate × Attitude −.13 −.44 **
Certainty
  Incremental R2 = .01**

Total R2 = .16
Total R2Adjust = .14

**p < .05. ***p < .01. ****p < .001.

discussion frequency (β = .28, p < .001) are significant predictors of opinion expression.
Contrary to our second hypothesis, we find no significant effect of attitude certainty. There
is also no effect of opinion climate.
In the last block of the regression model, however, there is a significant interaction
between opinion climate and attitude certainty (β = −.44, p < .05). The negative sign of this
interaction signals that an increase in attitude certainty leads to a decrease in the relation-
ship between opinion climate and opinion expression. It is crucial to note that attitude
certainty and opinion climate are uncorrelated (r = −.04, ns). Thus, there are no grounds to
argue that attitude certainty serves as a mediator (i.e., a majority climate leads to attitude
certainty, which in turn leads to opinion expression) or that opinion climate is a mediator
for the relationship between attitude certainty and opinion expression (i.e., attitude cer-
tainty leads individuals to believe they are in a majority, which in turn impacts opinion
expression). Thus, our first hypothesis can be confirmed.
In order to probe this interaction, the Johnson-Neyman technique was applied including
the same list of predictors (producing the same effects and explaining variance as block-
wise regression). The Johnson-Neyman technique indicates the regions of the moderator

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

Matthes et al. 783

Figure 1. Willingness of opinion expression as a factor of majority opinion perception and

attitude certainty (Study 1).

for which the relationship between opinion climate and opinion expression is significant or
not significant. When the value for attitude certainty is 1.60 or below, the effect of opinion
climate on opinion expression is significantly positive (p < .05). By contrast, when attitude
certainty is above 1.60, no significant effect can be observed. Interestingly, for very high
levels of attitude certainty, the sign of the effect even turns negative but does not reach
statistical significance (at a value of 5, b = −.14, p = .19). Finally, we tested whether opin-
ion climate moderates the relationship between attitude certainty and opinion expression.
In fact, only for a minority (p < .05) but not for a majority climate, attitude certainty is a
significant and positive predictor of the willingness to express one’s views (see Figure 1).
Finally, to test whether attitude extremity also serves as a moderator, we ran the same
model but exchanged attitude certainty for attitude extremity (for reasons of multicol-
linearity; see above). The results of this analysis produce neither a significant effect of
attitude extremity on opinion expression (β = .05, ns) nor a significant interaction of atti-
tude extremity and majority perception (β = −.05, ns).1

Discussion
The pattern of results found in our first study revealed clear evidence for our line of reason-
ing. Respondents with relatively uncertain attitudes were more likely to express their opin-
ions when they perceived a majority rather than a minority opinion climate. In contrast, for
individuals who were confident in their attitudes, opinion climate was not relevant to opin-
ion expression. However, no such effect was found when attitude extremity was used as a
moderator. It is also important to stress that there was no direct effect of opinion climate on
opinion expression. Thus, if the moderating role of attitude certainty were ignored, no
indication for a spiral of silence would be observed in the data. Surprisingly, and in con-
trast to Hypothesis 2, no main effect of attitude certainty was observed. This might be

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

784 Communication Research 37(6)

explained by the fact that the effect of attitude certainty is moderated by opinion climate.
Hence, attitude certainty does only enhance opinion expression when people feel they are
in the minority.
Some control variables were also significant. Older people were more likely to express
their opinions than younger people. This might be explained by a higher amount of politi-
cal sophistication and a higher issue-specific knowledge that older citizens have compared
to younger citizens. The positive effects of TV and newspaper use suggest that the news
media provide the content that is discussed (Moy et al., 2001; Noelle-Neumann, 1993;
Scheufele, 1999). The more people read, see, or hear about politics in the news, the more
likely it is that they will form opinions that can be expressed. The opposite is also plausi-
ble: The more people anticipate that they will be expressing their views, the more they turn
on the news or read the newspaper to prepare themselves for that. Finally, overall discus-
sion frequency was significantly related to opinion expression. The more frequently indi-
viduals discuss political issues, the more often they are given the opportunity to express
their views. Another explanation could be that frequent discussions of politics enhance
people’s subjective feeling of practice or experience in opinion expression.
Taken together, the findings of Study 1 confirm the hypothesized moderating role of
attitude certainty. However, some caution is warranted in drawing this conclusion. On the
theoretical side, we have argued that attitude certainty is the key variable in identifying the
hardcore. Indeed, attitude extremity did not moderate the relationship between climate and
opinion expression. However, we were not able to test the effects of other dimensions of
attitude strength, such as importance (see Krosnick et al., 1993). Thus, more evidence is
needed for the claim that attitude certainty—and not a related construct—explains the
hardcore phenomenon.
On the methodological side, the operationalization of majority climate was not opti-
mal. Respondents were split into two groups based on their own attitudes toward the EU
and their estimation of the likelihood of Switzerland joining the EU. Therefore, the
study needs to be replicated working with a more refined measure of climate percep-
tion. Moreover, the explained variance of the interaction effect was rather small, raising
questions about the robustness of this effect. As has often been noted (e.g., Kelly, Chase,
& Tucker, 1979), no single study can be considered definitive with respect to some
research issue or question. Nevertheless, replicating the findings of Study 1 with the
same list of predictors would lend more confidence to our line of reasoning. Therefore,
the aim of Study 2 was to replicate the findings of Study 1 in a different context using
more refined measures, and to assess perceived importance as an additional dimension
of attitude strength.

Study 2
Method

The second study was administered through a Web survey pertaining to the World Eco-
nomic Forum’s (WEF) annual meeting, a major global event that took place in Davos,

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

Matthes et al. 785

Switzerland, from January 23 to January 27, 2008. Respondents were solicited by placing
announcements on a variety of newsgroups, blogs, text, and banner ads on high-traffic
national newspaper Web sites, homepages of politicians and political activists, and on offi-
cial sites by the local state government. Special attention was paid to recruiting both oppo-
nents and proponents of the WEF. The survey was available online from January 23 to
February 14, 2008. In order to increase data quality, respondents who did not complete the
survey from beginning to end and respondents who completed the survey more quickly
than pilot testing suggested was possible were excluded (cutoff was 15 minutes; a begin-
ning and ending time stamp was provided for each record by the Web survey software).
Furthermore, an automatic Internet protocol check reduced the likelihood that an individ-
ual could respond to the survey more than once. The final sample included 1,096 respon-
dents (26% female, Mage = 41.33, SD = 17.26, range = 18 to 86 years). Respondents with
higher educational degrees were overrepresented (35% college degree, 27% high school
degree that is necessary to go to a college).

Measures
Willingness to express one’s opinion was measured by asking respondents whether they
had voiced their opinion about the WEF (a) in front of other people, (b) in front of close
friends, (c) in front of neighbors, and (d) in front of spouse/boyfriend/ girlfriend. These
variables were summed up to an index ranging from 0 to 4 (M = 1.76, SD = 1.26). Attitude
certainty was again measured with a standard item asking, “How certain are you in your
opinion about the WEF?”, ranging from 1 (very uncertain) to 5 (very certain; M = 3.92,
SD = 0.93). As alternative measures of attitude strength, we included issue importance and
attitude extremity. Again, extremity (M = 2.55, SD = 1.36) was operationalized as the abso-
lute value of the difference between participants’ own attitude (“How positive or negative
do you think about the World Economic Forum?”; 1 = do not agree at all, 10 = fully agree;
M = 4.62, SD = 3.23) and the scale’s midpoint. For perceived issue importance (see Kim
et al., 2004; Moy et al., 2001), participants responded to the item, “The issue is very impor-
tant to me personally” (1 = do not agree at all, 5 = fully agree; M = 2.74, SD = 1.36).
(Minority) opinion climate was assessed with the item, “Most people in Switzerland
think differently about the WEF compared to me,” ranging from 1 (strongly disagree) to 5
(strongly agree). The rationale behind this item was to assess opinion climate in a more
direct and precise way, rather than combine two measures as done in Study 1 and previous
research (e.g., Scheufele et al., 2001; Willnat et al., 2001). The content of this item refers
to the difference between own opinion and perceived public opinion. Thus, it expresses
exactly the same content as the combination of the two measures in Study 1. In order to
denote majority opinion climate as in Study 1, the item was recoded (high value = majority
climate; M = 3.23, SD = 1.06).
The same controls as in Study 1 were included. Similar to the first study, sex (2 =
female, 1 = male), age, highest education, and frequency (1 = very seldom to 5 = very often)
of TV use (M = 3.47, SD = 1.32) and newspaper use (M = 4.16, SD = .96) were measured.
Furthermore, attitudes toward the WEF were also controlled. Issue attitude was measured

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

786 Communication Research 37(6)

with one item (“How positive or negative do you think about the World Economic Forum?”;
see above), and general discussion frequency was gauged with the item, “How often do
you talk with family and friends about political issues?”(1 = never to 5 = very often; M =
4.40, SD = 0.76).

Data Analysis
Data analysis was similar to Study 1. The interaction terms (certainty × climate and impor-
tance × climate) served as the last block in a moderated regression. When an interaction
term was nonsignificant, it was left out of the final model (see Hayes, 2005). Since attitude
certainty and extremity were highly correlated (r = .48, p < .001) signaling problems of
multicollinearity, attitude extremity was exchanged for attitude certainty in an additional
analysis. Importance, however, correlated only moderately with both certainty (r = .19,
p < .001) and extremity (r = .15, p < .001), and was thus left in the model.

Results
Table 2 shows the results of Study 2. Among the statistical controls, only frequency of TV
news (β = .14, p < .001), issue attitude (β = −.20, p < .001), and overall discussion fre-
quency (β = .28, p < .001) exert a significant influence on opinion expression. The more
certain individuals are in their attitudes toward the WEF, the more they are willing to
express their views publicly (β = .18, p < .001). Also, issue importance has a significant
effect on opinion expression (β = .13, p < .001).
As in the previous study, there is no direct evidence for an unmoderated spiral of
silence: The perception of opinion climate is unrelated to opinion expression (β = .04, ns).
In the last block of the regression, however, a marginally significant interaction effect of
attitude certainty and opinion climate can be observed (β = −.29, p = .06). A graph of the
interaction is shown in Figure 2. The negative sign of the interaction term denotes that the
relationship between climate perception and opinion expression decreases when attitude
certainty increases. Because attitude certainty is not positively related to the perception of
a majority climate, it cannot be argued that attitude certainty causes a majority perception
or that a majority opinion climate leads to more certain attitudes. In fact, the correlation is
very small and surprisingly even negative (r = −.08). Thus, confirming our hypothesis,
attitude certainty moderates the relationship between opinion climate and opinion expres-
sion. It does not mediate this relationship, nor does it predict one’s perception of the opin-
ion climate.
The interaction term of importance and majority climate was not significant (β = −.04,
ns) and was left out of the model (see Hayes, 2005). In an additional analysis, we included
attitude extremity instead of attitude certainty. Like attitude certainty, attitude extremity is
significantly related to opinion expression (β = .21, p < .001). However, there is no interac-
tion between extremity and opinion climate (β = .13, ns).
Thus, only attitude certainty interacts with majority climate perception. In probing this
interaction with the Johnson-Neyman technique, it can be shown that opinion climate is a

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

Matthes et al. 787

Table 2. Regression Coefficients for Predicting Opinion Expression (Study 2)

Unstandardized Standardized
Predictors Coefficients Coefficients p Value
Block 1: Demographics
  Age −.00 −.06
  Sex .01 .00
  Education .02 .02
  Incremental R2 = .00
Block 2: Media use
  Frequency TV news use .12 .14 ****
  Frequency newspaper use −.09 −.07
  Incremental R2 = .02***
Block 3: Issue attitude and discussion
  Issue attitude −.08 −.20 ****
  Overall discussion frequency .46 .28 ****
  Incremental R2 = .12****
Block 4: Focal predictors
  Perceived majority climate .04 .04
  Attitude certainty .24 .18 ****
  Importance .13 .13 ****
  Incremental R2 = .05****
Block 5: Interaction
  Perceived Majority Climate × Attitude −.07 −.29 *
Certainty
  Incremental R2 = .01*

Total R2 = .15
Total R2Adjust = .13
*p = .06. **p < .05. ***p < .01. ****p < .001.

significant predictor of opinion expression for individuals with relatively low and moder-
ate attitude certainty. Up to an attitude certainty of 3.82, the relationship between both
variables is statistically significant (p < .05). When attitude certainty is higher than 3.82,
opinion climate has no effect on opinion expression. Finally, the other possible direction of
the interaction was probed: A significant main effect of attitude certainty on opinion
expression can be found across all levels of opinion climate (see Figure 2). However, this
effect is weaker (p < .05) for very high levels (i.e., scoring 5 on the scale) compared to low,
moderate, or high levels (p < .001) of majority climate perception.

Discussion
The aim of Study 2 was to replicate Study 1 with a more precise measurement of opinion
climate and with issue importance as another potential moderator. Results largely con-
firmed the findings of Study 1. Consumption of TV news and overall discussion frequency

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

788 Communication Research 37(6)

Figure 2. Willingness of opinion expression as a factor of majority opinion climate and

attitude certainty (Study 2).

were again related to opinion expression. In addition, having a negative issue attitude
toward the WEF predicted the expression of opinions. This might be explained by the fact
that being against something may involve negative emotions such as anger and threat.
These emotions increase evaluations of importance and concern (e.g., Marcus & McKuen,
1993), thereby enhancing the likelihood of joining discussions.
Also in line with Study 1, there was again no unmoderated spiral of silence: The minor-
ity opinion climate did not lead to a decrease in opinion expression per se. However, for
individuals with low or moderate attitude certainty, clear evidence for an influence of the
climate of public opinion was found. No moderated effects were found for issue impor-
tance and attitude extremity. These findings underscore the unique relevance of attitude
certainty to the relationship between opinion climate and opinion expression. In line with
our second hypothesis, there was a main effect of attitude certainty on opinion expression.
However, this effect was smaller for a minority compared to a majority climate.
Despite this compelling evidence for our argument, the results of both studies should
still be interpreted with caution. There are essentially three drawbacks. First, the perceived
opinion climate was measured by a combination of questions in Study 1 and with a single
item in Study 2. Although the results of both studies are consistent, the use of single items
to capture this key independent variable is troublesome.2 Thus, Study 3 aims to replicate
the findings of Studies 1 and 2 using a multiple item measure of opinion climate.
Second, the indicators we used in Studies 1 and 2 measured overall opinion expression,
not dangerous opinion expression (see Eveland & Hively, 2009; Eveland, Morey, & Hively,
2008). Dangerous opinion expression, that is, expressing an opinion to people who tend to

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

Matthes et al. 789

disagree with this opinion, is more closely at the heart of spiral of silence theory (see
Salmon & Neuwirth, 1990). It encapsulates the tendency of individuals to silence them-
selves in presence of a hostile climate. It can be expected that dangerous opinion expres-
sion should be strongly affected by a hostile opinion climate. In other words, when a person
holds a minority opinion, the likelihood of falling silent should be higher in a dangerous
setting compared to a safe setting. Thus, an indicator of dangerous discussion should pro-
vide a stricter test of spiral of silence theory.
Third, Studies 1 and 2 can potentially be criticized for their sample and sampling meth-
odology. Although Web-based surveys are largely accepted in the scientific community, it
is fair to say that representative telephone surveys working with random sampling provide
more reliable and more valid results. Thus, another aim of Study 3 was to replicate our
findings with a representative, national sample.

Study 3
Method

Data from the third study were collected as part of a representative survey on a national
referendum about the naturalization of Swiss immigrants. This issue deals with the appli-
cation process and decision procedures for immigrants who want to attain national citizen-
ship. The sample was recruited by random digit dialing (random quota) in June 2008 (N =
1,201, age M = 48.29, SD = 16.71; 50.6% female; nationality: German-speaking (78.4%)
and French-speaking Swiss residents). The questionnaire was programmed for an applica-
tion of Computer Assisted Telephone Interview. To ensure high data quality, the study was
conducted by a major global polling company that adopts internationally established poll-
ing standards.

Measures and Data Analysis

Respondents were asked about their opinions toward naturalizing immigrants. Opinion
climate was measured with an index of two items (α = .54, M = 3.14, SD = 0.94; 1 = do not
agree at all to 5 = fully agree; “My opinion on the issue of naturalization equals most
opinions I hear in my surroundings,” “In the place where I live, most people think the same
way about the issue of naturalization than I do”).3 Attitude certainty was again gauged by
asking how certain people felt in their opinion (1 = very uncertain to 5 = very certain; M =
4.25, SD = 0.95), and dangerous discussion was assessed by asking how often respondents
discussed this issue with people who have a different opinion (1 = very seldom to 5 = very
often; M = 2.21, SD = 1.20). Personal importance (see Kim et al., 2004; Moy et al., 2001)
was gauged with the item, “The issue of naturalization is very important to me personally”
(1 = do not agree at all to 5 = fully agree; M = 4.14, SD = 1.11). Attitude extremity (M =
1.45, SD = 0.92) was computed by taking the absolute value of the difference between
participants’ own attitude (see below) and the scale’s midpoint. This time, attitude certainty
and extremity were only moderately correlated (r = .29, p < .001), signaling no concerns
about multicollinearity. Both measures were thus left in the model.

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

790 Communication Research 37(6)

As controls, we included sex (2 = female, 1 = male), age, and education. Importance of

television (M = 3.83, SD = 1.24) and newspaper (M = 3.89, SD = 1.22) for being informed
about politics (1 = very unimportant to 5 = very important) and a measure of general dis-
cussion (average score of the amount of interpersonal communication with family/people
one lives with, spouse/partner, and friends; 1 = very seldom to 5 = very often; α = 71; M =
2.54, SD = 1.17) were measured. Issue attitude was measured with a reliable index of three
items (α = 72; M = 3.98, SD = 1.71, “I approve stricter naturalization rules in Switzerland,”
1= not agree at all to 10 = fully agree; “There are already too many foreigners in Switzer-
land”; and “Mass naturalizations must be stopped,” 1 = not agree at all to 5 = fully agree).
Data analysis was equivalent to the two previous studies.

Results
Several controls exert a significant effect on dangerous discussion (see Table 3). Sex (β =
−.12, p < .001), education (β = −.09, p < .01), and newspaper use (β = .08, p < .05) are
significant predictors. The strongest predictor of all controls, however, is the amount of
overall discussion that our respondents reported (β = .50, p < .001). There is no effect of
attitude certainty and opinion climate on dangerous opinion expression.
More importantly, however, there is again a significant interaction of attitude certainty
and opinion climate (β = −.29, p < .05). This is in line with our first hypothesis that attitude
certainty moderates the relationship between opinion climate and opinion expression.
Although attitude certainty is weakly related to the perception of a majority climate (r =
.06, p < .05), this correlation disappears if the same controls are included as in Table 3. It
follows that alternative explanations can be ruled out: Attitude certainty is neither a media-
tor between climate and opinion expression nor is it a predictor of opinion climate.
As Table 3 reveals, there is no effect of personal importance or attitude extremity on
dangerous discussion. The interactions of importance (β = −.09, ns) and extremity (β =
−.08, ns) with majority climate are not significant and thus left out of the model.
Hence, only attitude certainty moderates the effect of opinion climate; attitude extrem-
ity and importance do not. Results for probing and interpreting this interaction effect with
the Johnson-Neyman technique are as follows (see Figure 3): When attitude certainty is
low (below a value of 1.8), a significant effect (p < .05) of opinion climate perception on
dangerous discussion can be observed. Above that value, there is no relationship whatso-
ever. Interestingly, for very high values of attitude certainty, the sign of the relationship
turns negative but does not reach statistical significance. Finally, probing the interaction
reveals that there is a significant positive effect (p < .05) of attitude certainty on opinion
expression for low levels of perceived majority climate (values below 1.64).

Discussion
Study 3 shows that dangerous discussion depends on a variety of factors. It is not surpris-
ing that men prefer dangerous discussion more than do women, as men are more willing to
express dissent (Maslach, Santee, & Wade, 1987). Higher educated individuals are less

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

Matthes et al. 791

Table 3. Regression Coefficients for Predicting Dangerous Discussion (Study 3)

Unstandardized Standardized
Predictors Coefficients Coefficients p Value
Block 1: Demographics
  Age −.00 −.04
  Sex −.29 −.12 ***
  Education −.03 −.09 **
  Incremental R2 = .02****
Block 2: Media use
  Frequency TV news .04 .04
  Frequency newspaper use .08 .08 ***
  Incremental R2 = .01***
Block 3: Issue attitude and discussion
  Issue attitude .00 .01
  Overall discussion frequency .52 .50 ****
  Incremental R2 = .25****
Block 4: Focal predictors
  Perceived majority climate .00 .00
  Attitude certainty .02 .02
  Importance −.02 −.02
  Attitude extremity −.04 −.03
  Incremental R2 = .00
Block 5: Interaction
  Perceived Majority Climate × Attitude −.07 −.29 **
Certainty
  Incremental R2 = .002**

Total R2 = .28
Total R2Adjust = .27
**p < .05. ***p < .01. ****p < .001.

willing to express their opinions than less educated ones, possibly because less educated
people are more concerned about negative effects of immigration (e.g., fewer job opportu-
nities for Swiss citizens). As before, the importance of some news media can be explained
either by the fact that news media provide the information to engage in dangerous discus-
sion or by the argument that dangerous discussion increases news media use in an attempt
to stay informed. As in the previous two studies, overall discussion frequency is a strong
predictor: Individuals who generally discuss their opinions more than others do are also
more likely to join dangerous discussions.
More importantly, as in the previous studies, there is no unmoderated effect of opinion
climate on dangerous discussion. In fact, the Johnson-Neyman technique revealed that for
individuals who hold attitudes about the naturalization of immigrants with low certainty,
majority climate is a significant predictor of dangerous discussion. In other words, people
in the minority position are likely to withhold their opinions, whereas people who feel they
are in the majority are more likely to express their views. However, when attitude certainty

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

792 Communication Research 37(6)

Figure 3. Willingness of dangerous discussion as a factor of majority opinion climate and

attitude certainty (Study 3).

is moderate or high, no such relationship can be found. These individuals will always voice
their opinions no matter how friendly or hostile the opinion climate is. Attitude certainty,
extremity, and personal importance are unrelated to dangerous discussion, which points to
the unique moderating role of attitude certainty in this process.

General Discussion
Spiral of silence theory does not assume a simple relationship between opinion climate and
opinion expression. The notion of hardcore individuals embraces the idea that there are
some people for whom this relationship does not hold true. However, previous research has
treated this notion as an assumption and has not attempted to measure the hardcore phe-
nomenon empirically. In this article, we have proposed that attitude certainty is a key vari-
able in identifying hardcore individuals.
In three independent surveys, we have provided clear evidence for our first hypothesis
that the climate of opinion only determines opinion expression when individuals hold their
attitudes with weak or moderate attitude certainty. Importantly, the opinion climate had no
main effect on opinion expression. In other words, without including attitude certainty as
a moderating variable, no support for a spiral of silence would have been found in any of
the three studies. Furthermore, the moderating effect of attitude certainty was even present
after several other variables were controlled for, variables that have been shown to relate
to opinion expression in previous research.
All three surveys demonstrated that attitude certainty, but not attitude extremity or issue
importance, moderates the spiral of silence. In terms of spiral of silence theory, the findings
suggest that it is not enough to have an extreme attitude or to find the attitude object

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

Matthes et al. 793

important in order to voice one’s opinion in a hostile opinion climate. It is rather conviction
that one possesses the truth that can break a spiral of silence. This is consistent with find-
ings of previous studies that attitude certainty is correlated with but distinct from other
dimensions of attitude strength (Tormala & Rucker, 2007).
When looking at the differences across studies, it is striking that both the main effect of
attitude certainty and the main effect of issue attitude can only be found in Study 2. One expla-
nation could be that the WEF is a more polarized issue compared to the other two issues. For
example, there were public demonstrations and protests against the WEF. In contrast, being
against Switzerland joining the EU or applauding stricter naturalization rules did not result in
opinion expression as these two issues are clearly less polarized. These two issues do not refer
to events that invite scheduled and organized protest but are established issues that are always
on the political agenda to some extent. In this light, it is not surprising that people against the
WEF were more likely to express their views than people in favor of the WEF.
Similarly, the issues covered in Studies 1 and 3 somehow relate to referendums in
which people are given the opportunity to vote. Study 3 was conducted 8 weeks prior to a
national referendum about the naturalization of immigrants, and the bilateral relationship
between Switzerland and the EU has been an issue of past referendums. For the WEF,
however, people could voice their opinions in that debate, but they had no possibility at all
to change or stop the forum; they could not vote for something so they had to find other
ways to participate. This might have contributed to the observed effects of attitude cer-
tainty and issue opinion on opinion expression.
These results, however, are somewhat contradictory to the findings of others (e.g.,
Lasorsa, 1991) who observed a significant main effect of attitude certainty. Thus, there was
only mixed support for our second hypothesis. Given the differences in the studies’ context,
issue, and sampling, there could be many explanations for this. It should be stressed, how-
ever, that previous research did not test whether the observed effect of attitude certainty holds
true across all levels of opinion climate. The reason is that the interaction between opinion
climate and attitude certainty was not included in those models. In fact, even in Study 2 in
which a main effect of attitude certainty appeared, this effect was weaker with rising levels
of majority climate. Thus, more research is needed to explain these differences.

Theoretical Implications
In considering the broader implications of our findings, four things become apparent. First,
these results can be used to argue that future spiral of silence research should not only
observe the direct, main effect of opinion climate (or attitude certainty) but also focus on
moderation. This can have important consequences for our judgment about the status of
spiral of silence theory (Eveland, 1997). As stated above, without examining moderated
effects, scholars could conclude that there is no spiral of silence when finding no main
effects of climate perceptions. This conclusion, however, could possibly not be justified.
The notion of the hardcore in spiral of silence theory explicitly assumes a moderated effect.
Thus, when this notion is not taken into account, the theory is not being adequately tested.
Second, we have proposed an issue-specific measurement of the hardcore phenomenon,
whereby individuals can hold their attitudes with certainty on some issues but with less

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

794 Communication Research 37(6)

certainty on others. We argue that this is more appropriate for the highly issue-specific
phenomenon of opinion expression than measuring general (i.e., trait) certainty. Of course,
there are some traits that predict individuals’ tendencies to voice their opinions in a hostile
opinion climate. For example, Hayes et al. (2005) defined the willingness to self-censor as
“a person’s general reticence to express an opinion to an audience that is likely to disagree”
(p. 319). Hayes and colleagues demonstrated that the climate of opinion affects opinion
expression only among dispositional self-censors. In line with Hayes et al., we assume that
there are some individuals who will always voice their opinions when they face a minority
situation, no matter what the issue or context. Thus, the hardcore phenomenon may be both
general and issue specific. However, attitude certainty should explain more variance in
specific situations where the influence of general dispositional tendencies is weak. It can
also be expected that high self-censorers are especially likely to break the spiral of silence
when they hold very certain attitudes. Clearly, further research is needed to disentangle
trait and issue-specific influences on minority opinion expression.
Third, and closely related to this, the results of the present research suggest that the
hardcore people might be a larger group than commonly thought. Additional descriptive
analysis shows that approximately only one third of our samples had uncertain or moder-
ately certain attitudes. This, in fact, means that the spiral of silence is only relevant to a
rather small part of the public. In contrast, previous research, and especially Noelle-
Neumann (1993) herself, has assumed that the hardcore are a small minority in a society
and not actually a prevailing majority. Although we have to be very careful about the gen-
eralization of these descriptive findings to other issues or contexts, it may be worthwhile
to reconsider the argument that hardcore individuals are the exception to the rule, that is,
assuming a spiral of silence for all. At least in our studies, the majority of all respondents
chose to speak out although they were facing a hostile opinion climate. A spiral of silence
could only be observed for some.
Fourth, our findings also implicitly challenge Noelle-Neumann’s (1993) theoretical
argument that the spiral of silence only works for moral or value-laden issues. On the con-
trary, it can be speculated that value-laden issues are those issues that polarize opinions,
that is, those that lead to rather certain attitudes. Put differently, when an issue has a moral
element, there should be a higher likelihood that individuals get deeply invested in those
issues. This in turn would lead to high attitude certainty, which eventually would weaken
a spiral of silence. In preliminary support of this idea, Hornsey et al. (2003) found that
individuals with a strong moral basis for their attitudes were somewhat more likely to
express their minority than majority opinions (i.e., they exhibited counterconformity). One
explanation is that these individuals—relative to their counterparts with a weak moral
basis for their attitudes—felt more certain that their attitudes were correct, thus causing
them to voice their opinions more freely. In other words, attitude certainty may mediate the
relationship between moral basis for one’s attitude and (lack of) spiral of silence effects.
However, issues that lack any kind of moral component are also unlikely to elicit a spiral
of silence, simply because people might not even care what others think. Following this
line of reasoning, one could argue that a spiral of silence is most likely to be found for
moderately value-laden issues. Again, future research is needed to test these speculations.

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

Matthes et al. 795

Limitations
There are some drawbacks of our studies that are worthy of careful consideration. To begin
with, although we included the same list of controls in all three studies, there were slight
differences in item wordings for those controls. Nevertheless, the effects were largely con-
sistent across all three studies.
Another drawback is that we used single item measures for some variables, including:
(a) attitude certainty, (b), opinion climate, and (c) opinion expression. To measure attitude
certainty, we used a standard item that has been established in previous research and by
leading experts in research on attitude certainty (e.g., Fazio & Zanna, 1978; Petrocelli
et al., 2007; Tormala & Petty, 2002). Because this item has been validated in such a large
body of scholarly work and we are not aware of an equally prevalent measure of attitude
certainty, we included it in the present studies.
When it comes to perceived opinion climate, however, there is no equally established
single-item measure available in the literature. Many studies have used single items, but
most of them are rather different from one another (e.g., Kim et al., 2004; Moy et al.,
2001). In Study 1, we operationalized opinion climate by combining two measures as has
been done in numerous spiral of silence studies (Neuwirth, 2000; Scheufele et al., 2001;
Willnat et al., 2001). Study 2 was designed to replicate Study 1 using a direct assessment
of opinion climate. Because the single-item operationalizations in both studies are poten-
tially troublesome, we used a two-item measure of opinion climate in Study 3. Despite
these differences in operationalization, however, results were consistent across all three
studies and provide convergent evidence in support of our hypothesis.
Finally, the key dependent variable was operationalized as an index based on several
items in Studies 1 and 2, rather than by using single items tapping hypothetical scenarios.
This is an established procedure in spiral of silence research (e.g., Perry & Gonzenbach,
2000; Petrič & Pinter, 2002; Salmon & Oshagan, 1990). In Study 3, a single-item measure
was used to capture dangerous discussion. More refined measures should be used for that
construct in future studies.
Taken together, while there are problems with single-item measures in some studies, all
key dependent and independent variables were appropriately measured in at least one
study. While a particular study can be criticized for certain operationalizations, the cumula-
tive evidence across all three studies lends credence to our theorizing.

Future Research
Although we were able to observe the very same effect in three independent studies, work-
ing with different issues and sometimes different measures, more research is needed to
prove the causal influence of attitude certainty. One useful future direction would be to
conduct a panel study that demonstrates the moderating influence of attitude certainty over
several points in time. This would allow us to actually speak of true spirals (for the same
argument, see Neuwirth et al., 2007).

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

796 Communication Research 37(6)

Related to this, we have shown that attitude certainty serves as a moderator, and not as
a mediator, of the relationship between climate perceptions and opinion expression. All
three of our studies investigated existing political issues. For such issues, on which people
are likely to have already established their attitudes, the climate of opinion might be less
important in explaining attitude certainty. For novel issues (e.g., an unfamiliar or fictitious
policy), however, it is indeed plausible that a majority opinion climate can increase attitude
certainty (see Eaton, Majka, & Visser, 2008). The reason is that people may feel a stronger
need to search the opinion climate for cues about what to think on novel (compared to
existing) issues. Thus, treating attitude certainty both as a moderator of majority opinion
climate on opinion expression and as a consequence of majority opinion climate in future
studies seems to be a worthwhile effort.
Finally, the construct of attitude certainty itself needs further elaboration. Recently,
Petrocelli et al. (2007) demonstrated that attitude certainty can be conceptualized and
empirically distinguished in terms of attitude clarity (knowing what one’s attitude is) and
attitude correctness (knowing that one’s attitude is correct). Our understanding of attitude
certainty is most consistent with the notion of attitude correctness. Thus, distinguishing
attitude correctness from attitude clarity when investigating moderated effects in the spiral
of silence could prove useful in future studies.

Conclusion
The present article suggests that spiral of silence theory can be enriched by giving more
weight to the idea of hardcore individuals who are willing to express dissenting viewpoints.
We suggest that attitude certainty as an issue specific variable can capture this idea. In three
studies, we have shown that a hostile opinion climate impacts opinion expression only among
those who are not very certain about their attitudes. Establishing this boundary condition can
potentially contribute to the detection of opinion spirals in future empirical research.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the authorship and/or
publication of this article.

Funding
The authors disclosed receipt of the following financial support for the research and/or au-
thorship of this article: The research conducted in Study 3 of this article was supported by a
grant from the Swiss National Science Foundation as a part of the National Center of Compe-
tence in Research (NCCR) “Challenges to Democracy in the 21st Century.”

Notes

1. In an additional test, we explored whether the moderating role of attitude certainty is issue
specific, or if there is a situation-independent, trait-like variance in attitude certainty that can
account for the results we observed. The rationale for this test was that there may be some

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

Matthes et al. 797

people who are always certain of their opinions. In other words, when they hold opinions,
they hold them with certainty. To account for this variance, we added the personality trait
need to evaluate (NTE; Jarvis & Petty, 1996) as a focal independent variable and moderator.
NTE has been shown to be a very strong trait component of certainty (Britt et al., 2009). NTE
was measured in Study 1 with four items (α = .74; 5-point scale, M = 3.20, SD = 0.79; “It’s
very important for me to hold strong opinions,” “I like to have strong opinions even when I
am not personally involved,” “I form opinions about everything,” and “I would rather have a
strong opinion than no opinion at all”). Items were selected based on their high factor load-
ings in Jarvis and Petty (1996). As a result, NTE has a main effect on opinion expression
(β = .15, p < .001), but there is no interaction with opinion climate (β = −.23, p = .35). Thus,
the issue specific variance in attitude certainty, not the general tendency to hold opinions with
certainty across several issues, is responsible for the moderator effect we have observed, not
the general tendency to hold opinions with certainty across several issues.
2. Attitude certainty was also measured with a single item. However, this item is a standard
measure in psychological research and as such is generally accepted and widely used (Fazio
& Zanna, 1978; Petrocelli et al., 2007; Tormala & Petty, 2002).
3. In contrast to Studies 1 and 2, this measurement refers to reference groups rather than the opin-
ion climate present in the whole country. The measure was chosen because it is more appropri-
ate to explain the dependent variable dangerous discussion compared to opinion expression in
various public settings. Dangerous discussion is likely to happen in people’s close surround-
ings, as it is rather unlikely that people start to argue with complete strangers who hold dif-
ferent views. Thus, one’s immediate opinion climate is more relevant to explaining dangerous
discussion than is the opinion climate prevalent in the whole country (e.g., Oshagan, 1996).

References
Baldassare, M., & Katz, C. (1996). Measures of attitude strength as predictors of willingness to speak to the
media. Journalism & Mass Communication Quarterly, 73, 147-158.
Bassili, J. N. (1996). Meta-judgmental versus operative indexes of psychological attributes: The case of
measures of attitude strength. Journal of Personality and Social Psychology, 71, 637-653.
Bauer, D. J., & Curran, P. J. (2005). Probing interactions in fixed and multilevel regression: Inferential and
graphical techniques. Multivariate Behavioral Research, 40, 373-400.
Bizer, G. Y., Tormala, Z. L., Rucker, D. D., & Petty, R. E. (2006). Memory-based versus on-line processing:
Implications for attitude strength. Journal of Experimental Social Psychology, 42, 646-653.
Britt, T. W., Millard, M. R., Sundareswaran, P. T., & Moore, D. (2009). Personality variables predict
strength-related attitude dimensions across objects. Journal of Personality, 77, 859-882.
Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. New York: Harcourt Brace Jovanovich.
Eaton, A. A., Majka, E. A., & Visser, P. S. (2008). Emerging perspectives on the structure and function of
attitude strength. European Review of Social Psychology, 19, 165-201.
Eveland, W. P. (1997). Interactions and nonlinearity in mass communication: Connecting theory and meth-
odology. Journalism & Mass Communication Quarterly, 74, 400-416.
Eveland, W. P. Jr., & Hively, M. H. (2009). Political discussion frequency, network size, and “heterogene-
ity” of discussion as predictors of political knowledge and participation. Journal of Communication.
59, 205-224.

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

798 Communication Research 37(6)

Eveland, W. P., Morey, A., & Hively, M. (2008). Distinguishing dimensions of political discussion using
demographic, media use, political and personality variables. Paper presented at the annual meeting of
the AEJMC, Chicago.
Fazio, R. H., & Zanna, M. P. (1978). Attitudinal qualities relating to the strength of the attitude-behavior
relationship. Journal of Experimental Social Psychology, 14, 398-408.
Glynn, C. J, Hayes, A. F., & Shanahan, J. (1997). Perceived support for one’s opinions and willingness to
speak out: A meta-analysis of survey studies on the “spiral of silence.” Public Opinion Quarterly, 61,
452-463.
Glynn, C. J, & McLeod, J. M. (1984). Public opinion du jour: An examination of the spiral of silence. Public
Opinion Quarterly, 48, 731-740.
Haddock, G., Rothman, A. J., Reber, R., & Schwarz, N. (1999). Forming judgments of attitude certainty,
intensity, and importance: The role of subjective experiences. Personality and Social Psychology Bul-
letin, 25, 771-782.
Hayes, A. F. (2005). Statistical methods for communication science. Mahwah, NJ: Lawrence Erlbaum.
Hayes, A. F., Glynn, C. J., & Shanahan, J. (2005). Willingness to self-censor: A construct and measurement
tool for public opinion research. International Journal of Public Opinion Research, 17, 298-323.
Hayes, A. F., & Matthes, J. (2009). Computational procedures for probing interactions in OLS and logistic
regression: SPSS and SAS implementations. Behavior Research Methods, 41, 924-936.
Hayes, A. F., Scheufele, D. A., & Huge, M. E. (2006). Nonparticipation as self-censorship: Publicly observ-
able political activity in a polarized opinion climate. Political Behavior, 28, 259-283.
Ho, S. S., & McLeod, D. M. (2008). Social-psychological influences on opinion expression in face-to-face
and computer-mediated communication. Communication Research, 35, 190-207.
Hornsey, M. J., Majkut, L., Terry, D. J., & McKimmie, B. M. (2003). On being loud and proud: Non-
conformity and counter-conformity to group norms. British Journal of Social Psychology, 42, 319-335.
Jarvis, W. B. G., & Petty, R. E. (1996). The need to evaluate. Journal of Personality and Social Psychology,
70, 172-194.
Kelly, C. W., Chase, L. J., & Tucker, K. (1979). Replication in experimental communication research: An
analysis. Human Communication Research, 5, 338-342.
Kim, S., Han, M., Shanahan, J., & Berdayes, V. (2004). Talking on “sunshine” in North Korea: A test of the
spiral of silence as a theory of powerful mass media. International Journal of Public Opinion Research,
16, 39-62.
Krosnick, J. A., Boninger, D. S., Chuang, J. C., Berent, M. K., & Carnot, C. G. (1993). Attitude strength:
One construct or many related constructs? Journal of Personality and Social Psychology, 65, 1132-1151.
Krosnick, J. A., & Petty, R. E. (1995). Attitude strength: An overview. In R. E. Petty & J. A. Krosnick (Eds.),
Attitude strength: Antecedents and consequences (pp. 1-24). Hillsdale, NJ: Lawrence Erlbaum.
Krosnick, J. A., & Schumann, H. (1988). Attitude intensity, importance, and susceptibility to response
effects. Journal of Personality and Social Psychology, 54, 940-952.
Lasorsa, D. L. (1991). Political outspokenness: Factors working against the spiral of silence. Journalism
Quarterly, 68, 131-139.
Lavine, H., Borgida, E., & Sullivan, J. L. (2000). On the relationship between attitude involvement and atti-
tude accessibility: Toward a cognitive-motivational model of political information processing. Political
Psychology, 21, 81-106.

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

Matthes et al. 799

Mackie, D. M., & Asuncion, A. G. (1990). On-line and memory-based modification of attitudes: Determi-
nants of message recall–attitude change correspondence. Journal of Personality and Social Psychol-
ogy, 59, 5-16.
Marcus, G. E., & McKuen, M. (1993). Anxiety, enthusiasm, and the vote. American Political Science
Review, 87, 688-701.
Maslach, C., Santee, R. T., & Wade, C. (1987). Individuation, gender role, and dissent: Personality media-
tors of situational forces. Journal of Personality and Social Psychology, 53, 1088-1093.
Matthes, J., Wirth, W., & Schemer, C. (2007). Measuring the unmeasurable? Toward operationalizing
on-line and memory-based political judgments in surveys. International Journal of Public Opinion
Research, 19, 247-257.
McDonald, D. G., Glynn, C. J., Kim, S.-H., & Otsman, R. E. (2001). The spiral of silence in the 1948 presi-
dential election. Communication Research, 28, 139-155.
Miller, D. T., & Morrison, K. R. (2009). Expressing deviant opinions: Believing you are in the majority
helps. Journal of Experimental Social Psychology, 45, 740-747.
Moy, P., Domke, D., & Stamm, K. (2001). The spiral of silence and public opinion on affirmative action.
Journalism & Mass Communication Quarterly, 78, 7-25.
Neuwirth, K. (2000). Testing the spiral of silence model: The case of Mexico. International Journal of
Public Opinion Research, 12, 138-159.
Neuwirth, K., Frederick, E., & Mayo, C. (2007). The spiral of silence and fear of isolation. Journal of Com-
munication, 57, 450-468.
Neuwirth, L., & Frederick, E. (2004). Peer and social influence on opinion expression: Combining the theo-
ries of planned behavior and the spiral of silence. Communication Research, 31, 669-703.
Noelle-Neumann, E. (1974). The spiral of silence: A theory of public opinion. Journal of Communication,
24, 43-51.
Noelle-Neumann, E. (1993). The spiral of silence: Public opinion—Our social skin (2nd ed.). Chicago:
University of Chicago Press.
Oshagan, H. (1996). Reference group influence on opinion expression. International Journal of Public
Opinion Research, 8, 335-354.
Pan, Z., Shen, L., Paek, H., & Sun, Y. (2006). Mobilizing political talk in the 2000 presidential campaign.
An examination of campaign effects in a deliberative framework. Communication Research, 33, 1-31.
Perry, S. D., & Gonzenbach, W. J. (2000). Inhibiting speech through exemplar distribution: Can we predict
a spiral of silence? Journal of Broadcasting and Electronic Media, 44, 268-281.
Peter, J. (2004). Our long “return to the concept of powerful mass media”—A cross-national comparative
investigation of the effects of consonant media coverage. International Journal of Public Opinion
Research, 16, 144-168.
Petrič, G., & Pinter, A. (2002). From social perception to public expression of opinion: A structural equa-
tion modeling approach to the spiral of silence. International Journal of Public Opinion Research, 14,
37-53.
Petrocelli, J. V., Tormala, Z. L., & Rucker, D. D. (2007). Unpacking attitude certainty: Attitude clarity and
attitude correctness. Journal of Personality and Social Psychology, 92, 30-41.
Preacher, K. J., Rucker, D. D., & Hayes, A. F. (2007). Addressing moderated mediation hypotheses: Theory,
methods, and prescriptions. Multivariate Behavioral Research, 42, 185-227.

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012

800 Communication Research 37(6)

Prislin, R. (1996). Attitude stability and attitude strength: One is enough to make it stable. European Jour-
nal of Social Psychology, 26, 447-477.
Salmon, C. T., & Neuwirth, K. (1990). Perceptions of opinion “climates” and willingness to discuss the
issue of abortion. Journalism Quarterly, 67, 567-577.
Salmon, C. T., & Oshagan, H. (1990). Community size, perceptions of majority opinion, and opinion
expression. In L. A. Grunig, & J. E. Grunig (Eds.), Public relations research annual (Vol. 2, pp. 157-171).
Hillsdale, NJ: Lawrence Erlbaum.
Scheufele, D. A. (1999). Deliberation or dispute? An exploratory study examining dimensions of public
opinion. International Journal of Public Opinion Research, 11, 25-58.
Scheufele, D. A., & Moy, P. (2001). Twenty-five years of the spiral of silence: A conceptual review and
empirical outlook. International Journal of Public Opinion Research, 12, 3-28.
Scheufele, D. A., Shanahan, J. E., & Lee, E. (2001). Real talk: Manipulating the dependent variable in spiral
of silence research. Communication Research, 28, 304-324.
Shamir, J. (1997). Speaking up and silencing out in face of a changing climate of opinion. Journalism &
Mass Communication Quarterly, 74, 602-614.
Smith, S. M., Fabrigar, L. R., MacDougall, B. L., & Wiesenthal, N. L. (2008). The role of amount, cogni-
tive elaboration, and structural consistency of attitude-relevant knowledge in the formation of attitude
certainty. European Journal of Social Psychology, 38, 280-295.
Tormala, Z. L., & Petty, R. E. (2001). On-line versus memory-based processing: The role of “need to evalu-
ate” in person perception. Personality and Social Psychology Bulletin, 27, 1599-1622.
Tormala, Z. L., & Petty, R. E. (2002). What doesn’t kill me makes me stronger: The effects of resisting
persuasion on attitude certainty. Journal of Personality and Social Psychology, 83, 1298-1313.
Tormala, Z. L., & Rucker, D. D. (2007). Attitude certainty: A review of past findings and emerging perspec-
tives. Social and Personality Psychology Compass, 1, 469-492.
Willnat, L., Lee, W., & Detenber, B. H. (2002). Individual-level predictors of public outspokenness: A test
of the spiral of silence theory in Singapore. International Journal of Public Opinion Research, 14,
391-412.

Bios
Jörg Matthes (PhD, University of Zurich, 2007) is an assistant professor at the Institute of
Mass Communication and Media Research, University of Zurich. His research focuses on
public opinion formation, media effects, and empirical methods.

Kimberly Rios Morrison (PhD, Stanford University, 2008) is an assistant professor in the
School of Communication at the Ohio State University. Her research focuses on the roles of the
self-concept and social identity in public opinion expression and perception.

Christian Schemer (PhD, University of Zurich, 2009) is a postdoctural student at the Institute
of Mass Communication and Media Research, University of Zurich. His research focuses on
public opinion formation, media and emotions, and empirical methods.

Downloaded from crx.sagepub.com by Leida Ruvina on September 17, 2012