Heuristic Value of Opinionated News

The Heuristic Value of Opinions in the Market for News
Marcel Garz
Hamburg Media School and University of Hamburg
April 28, 2015
Address for correspondence: Marcel Garz, Hamburg Media School, Finkenau 35,
22081 Hamburg, Germany. E-mail: [email protected]. I thank
Christian Kolmer, Sven Meyer, Armin Rott, Johannes Schneller, and Matthias
Vollbracht for their support in obtaining and compiling the data, as well as
seminar participants in Hamburg and Esbjerg for helpful comment
1
Abstract
This study investigates the variation in the supply of and demand for opinionated
news. Due to different predispositions regarding systematic and heuristic
information processing, news consumers have varying preferences over
opinionated and objective reports. To verify the theoretical argument, I combine
content data on the news output of leading German media and survey information
from news consumers. The findings indicate consumption patterns that largely
correspond with the information-processing predispositions. I conclude that the
demand side of the news market determines the amount (more or less) and kind
(one-sided or ambivalent) of opinionation.
Keywords: news; opinionation; information processing; media bias
JEL classification: D11; D12; D80; L82
2
1. Introduction
Opinions are an essential part of the news. Often, viewpoints are not restricted to
editorials or columns, but can be found in purportedly objective forms of
journalism too. Among many other explanations, opinionated reporting is a
consequence of the complexity of world events. Journalistic work allows nonexperts to comprehend intricate issues, for example in politics and economics.
Giving opinion – that is, exercising judgment or quoting the views of others
(Benson & Hallin, 2007; Feldman, 2011a, 2011b) – is one journalistic method of
simplification. An external point of view serves as a cognitive shortcut if it
relieves the recipients from gathering and analyzing the underlying information to
form an own opinion. If certain conditions are met, judgments in the news thus
have a heuristic value for consumers.
Despite its role in audience persuasion and polarization (e.g., DellaVigna
& Gentzkow, 2010; Prior, 2013; Boukes et al., 2014), little is known about the use
of opinion as a tool of simplification. Why do media outlets vary in their
provision of opinionated versus objective news? Do recipients have different
preferences regarding the consumption of opinionated news, and if so, why?
One answer to these questions lies in the heuristic value of opinions. Due
to biological, evolutionary, and socio-cultural factors, for instance, some people
are predisposed to make more use of cognitive shortcuts than others. For example,
gender differences in information processing are well researched in the
psychology and consumer literature (for a recent review, see Meyers-Levy &
Loken, 2015): Women are known as comprehensive processors, whereas men tend
to be heuristic ones. As a consequence, male recipients can be expected to have a
greater demand for opinionated reports than female news consumers. Based on
information-processing theory, I derive further hypotheses about preferences over
opinionated and objective news. In particular, differences are hypothesized with
3
respect to the individual’s sex and gender, age, cognitive abilities, religion, and
socialization.
The empirical part of this study focuses on the German market for national
print news in the time from October 2001 to April 2007. Due to its corporatist
journalistic tradition, Germany is an ideal candidate for this evaluation: Tight
connections between the media, political parties, and interest groups favor
partisan and opinionated reporting (Esser & Umbricht, 2013). To verify the abovementioned hypotheses, I combine survey data from 70,292 news consumers with
information obtained from analyzing 119,838 political and economic reports of
the leading German print media. On this basis, I construct an indicator that
measures the respondents’ personal level of consumption of opinionated reports.
This indicator is then regressed on various explanatory variables that capture
individual information-processing predispositions.
The results indicate consumption patterns that largely correspond with the
information-processing predispositions: Characteristics that are associated with
systematic information processing increase the consumption of objective or
balanced news, whereas predispositions towards heuristic strategies favor the
consumption of news with one-sided opinions. In large parts, this relationship can
be interpreted causally, because people’s predispositions are based on
characteristics that are usually fixed (sex, gender, location of socialization), or
only change seldom or gradually (religion, cognitive abilities).
This study is closely related to the literature on demand-driven news
coverage and media bias (Mullainathan & Shleifer, 2005; Gentzkow & Shapiro,
2006, 2010; Chan & Suen, 2008). According to this literature, profit-maximizing
media companies slant their news coverage towards the beliefs of the recipients.
Media bias arises because there is a demand for the confirmation of views. This
study provides evidence that people’s information-processing predispositions are
another driver of opinionation and therefore media bias.
4
The next section provides the theoretical background and the hypotheses.
Afterwards, I describe the data and the empirical strategy. The estimation results
are discussed before the last section concludes.
2. Theory
2.1 Dual-Process Models
In social and cognitive psychology, dual-process models distinguish two modes of
human information processing: the systematic and the heuristic one (e.g.,
Chaiken, 1980; Chaiken & Trope, 1999). The former mode is characterized by
analytic thinking and effortful processing of all available information. In contrast,
the heuristic mode is based on limited information and the use of cues, rules of
thumb, and schemata that depend on past observations and experiences. Cognitive
shortcuts allow people to form opinions with little knowledge and effort. In
general, individuals tend to employ a heuristic approach when information costs
are high, and a systematic one when reliability is important. Both modes are ideal
constructs, because in reality there is a continuum of processing strategies.
2.2 Opinions as Cognitive Shortcuts
The heuristic value of certain kinds of information has been widely
acknowledged, for instance, in political decision making. Prominent examples are
a candidate’s party affiliation, ideology, and appearance, as well as endorsements
and poll results (Lau & Redlawsk, 2001). By investigating headlines as cognitive
shortcuts, Andrew (2007) highlights the role of news media in providing cues.
Opinions in news coverage can serve as cognitive shortcuts too. Although
professional journalism usually emphasizes the importance of objectivity,
opinions are not restricted to editorials or columns. Communications research
shows that news coverage in the US, Germany, and other countries has been
changing in the last decades. Instead of reporting mere events and facts,
5
journalists increasingly offer context, analysis, interpretation, and judgment
(Schudson, 1978; Wilke & Reinemann, 2001; Barnhurst, 2003; Fink & Schudson,
2014). Among other explanations, this development reflects the growing
complexity of many topics. As “makers of meaning”, journalists help their
audiences to understand complicated issues (Barnhurst & Mutz, 1997). Impartial
reporting often involves the provision of the whole spectrum of viewpoints, which
makes it difficult for people to comprehend (Graber, 2001; Patterson & Seib,
2005), especially in the case of political and economic topics. Judgments can
reduce the cognitive effort in processing information though, which facilitates the
formation of own opinions (Dalton, Beck, & Huckfeldt, 1998; Smith & Searles,
2013).
However, only reports that provide a clear, one-sided viewpoint can be
assumed to be of additional heuristic value. Balanced or ambivalent news
coverage – or two-sided messages in the phrasing of Zaller (1992) – requires an
increase in cognitive effort, because the recipients have to analyze contradicting
information on their own. The following hypotheses about the consumption of
opinionated news thus refer to the case of one-sided, unambiguous viewpoints.
2.3 Hypotheses
Hypothesis 1 (H1): Men consume more opinionated news than women.
Previous research shows that social and cultural gender roles influence
personality traits, which in turn cause differences in the way men and women
process information (Meyers-Levy, 1988; Meyers-Levy & Maheswaran, 1991;
Meyers-Levy & Sternthal, 1991; Darley & Smith, 1995; Laroche et al., 2000;
Meyers-Levy & Loken, 2015). Male gender roles are often characterized by
assertiveness and goal orientation. In contrast, female roles exhibit tendencies
towards interpersonal affiliation and communal objectives. These role differences
predispose women to be more sensitive to socio-emotional content and subtle
6
cues, which involves the comprehensive processing of information and low
thresholds of cognitive elaboration. Male gender roles, however, often cause men
to be selective processors that focus on self-relevant information and readily
available cues. Moreover, biological and evolutionary factors cause differences in
information processing to prevail, even when traditional gender roles dissolve in
modern societies (Buss, 1995; Putrevu, 2001): Several tasks that are associated
with childbirth and raising children require the ability to interpret subtle messages
and information that pertains to others. Similar to other areas of consumption,
sex- and gender-related predispositions should be detectable in the market for
news.
Hypothesis 2 (H2): The consumption of opinionated news increases with age.
Certain physical and mental abilities decline when people get older, which
often involves the speed at which the central nervous system processes
information. Older people compensate for the loss of processing speed by making
more use of their experience to employ cognitive shortcuts (Craik & Byrd, 1982;
Reder, Wible, & Martin, 1986). Previous research confirms the age-related shift
from effortful, systematic processing towards schema-based, heuristic strategies,
for example in the context of consumption decisions (Yoon, 1997; Yoon, Cole, &
Lee, 2009) and political evaluations (Riggle & Johnson, 1996).
Hypothesis 3 (H3): The consumption of opinionated news decreases with
cognitive abilities.
Cognitive abilities allow individuals to comprehend complex issues and
nuanced news coverage. According to the psychological concept of the need for
cognition (e.g., Cacioppo & Petty, 1982; Cacioppo et al., 1996), these abilities
enable people to process information systematically. Because engaging in
effortful, comprehensive thinking can provide competence feedback and social
7
reinforcement, people with high cognitive skills also develop the motivation for
mental elaboration. Cognitive abilities thus influence individual tendencies
towards systematic or heuristic information processing, which in turn shape
preferences over opinionated and objective news.
The concept of political sophistication – which refers to the number, range,
and degree of organization of a person’s political cognitions (Luskin, 1987) –
further supports the hypothesis. Politically sophisticated people are assumed to
possess high cognitive skills. Not only do these people consume more political
information than less sophisticated individuals, they also process it differently
(Rhee & Cappella, 1997). In particular, less sophisticated people more often rely
on cognitive shortcuts to form opinions than sophisticates (Kam, 2005; Boudreau
& McCubbins, 2010).
Hypothesis 4a (H4a): People with a religious affiliation consume more
opinionated news than people without a religious affiliation.
A religion constitutes a system of values, standards, and principles,
according to which a believer thinks and acts. Such a system provides mental
schema that have an impact on the cognitions of a believer (Lau, 1989). The
nature of religious values and principles often causes believers to adhere to higher
levels of conservative, authoritarian, and dogmatic beliefs than non-believers. As
a consequence, religious people tend to be more close-minded than non-religious
ones (Delener, 1990, 1994; McDaniel & Burnett, 1990). In contrast to openminded thinking – which involves the active search for and evaluation of
competing viewpoints – close-mindedness is usually associated with selective
processing of information and relying on heuristics. Thus people with and without
a religious affiliation likely differ in their consumption of opinionated news.
8
Hypothesis 4b (H4b): Catholics consume more opinionated news than Protestants.
The two largest religious groups in Germany – Catholics and Protestants –
exhibit differences that influence consumption behavior too. In Catholic families,
for example, patriarchal and authoritarian structures are more common than in
Protestant ones. Moreover, the Catholic Church mediates the believer’s
relationship to God to a larger extent than in Protestantism. Accordingly,
Catholics tend to be more traditional, dogmatic, and obedient to existing order
(Delener, 1990, 1994; McDaniel & Burnett, 1990). Due to these differences,
Protestants usually display greater levels of reliance on own judgments, and they
more readily engage in information-seeking activities (Choi, 2010). Compared to
Catholicism, Protestantism can thus be expected to predispose people towards
systematic information processing.
Hypothesis 5 (H5): East Germans consume more opinionated news than West
Germans.
During its separation after World War II, Germany was governed by two
different political systems, which has been affecting people’s attitudes and
behavior regarding information gathering and processing. With its market-based
approach, the West German democracy encouraged its citizens to live
individualist and self-responsible lives. In contrast, the communist East Germany
was characterized by a far-reaching social security system, guaranteed
employment, and the integration of its citizens into mass organizations. Moreover,
East Germans were subject to constant surveillance by the Ministry of State
Security, and subversive or dissident behavior was sanctioned severely. The
internalization of the system’s norms has been causing East Germans to exhibit
greater levels of obedience to authorities than West Germans (Mummert &
Schneider, 2001; Torgeler, 2003). Empirical research shows that east-west
differences in attitudes and preferences persist, even decades after reunification.
9
For example, East Germans are still more open to state intervention (Alesina &
Fuchs-Schündeln, 2007; Kuhn, 2012) and display lower levels of self-reliance and
entrepreneurial spirit (Bauernschuster et al., 2012).
In addition, East and West Germans experienced different kinds of news
coverage. In West Germany, freedom of speech and a free press facilitated the
provision of multiple opinions and balanced news. In contrast, the widely statecontrolled, East German press often provided only one-sided viewpoints. It is
likely that these reporting differences also have a persistent influence on people’s
news preferences. In this case – and considering the comparatively strong
predisposition towards belief in authority – it can be expected that the heuristic
value of unambiguously opinionated news is larger for East than for West
Germans.
3. Data
3.1 Sources and Sampling
To evaluate the hypotheses, I compile data from consumption surveys and media
content analyses.
The survey data come from Allensbach Institute, one of Germany’s most
important opinion and market research organizations. This institute conducts the
Allensbach Media Market Analysis (AWA, see www.ifdallensbach.de/awa/service/english/overview.html), which features representative
surveys of the German-speaking, resident population age 14 and over. Among
other things, the data contain information about demographic characteristics,
attitudes, and consumption habits. The six waves from 2002 to 2007 are available
for the purpose of this study. Each year’s wave comprises about 14,000 face-toface interviews, conducted between October of the previous year and April of the
respective year. The whole sample includes data from 85,757 interviews.
10
Information about the news output of German print media comes from
Media Tenor International (www.mediatenor.com). This media research institute
specializes on content analyses of political and economic news. The analyses are
conducted by trained experts who follow consistent instructions in codebooks;
their remuneration is based on inter-coder reliability and validity. The data used in
this study comprise the daily newspapers Frankfurter Allgemeine Zeitung,
Süddeutsche Zeitung, Die Welt, Frankfurter Rundschau; the daily tabloid Bild; the
weekly newspaper Die Zeit; and the weekly news magazines Spiegel, Stern, and
Focus. This selection includes all leading print media of national significance,
except for Financial Times Deutschland and Handelsblatt. I exclude these two
outlets because of the lack of consistent data. For the other nine outlets, the
sample includes all news on politics and economics, which amounts to 119,838
reports.
3.2 An Indicator of the Individual Consumption of Opinionated News
The AWA asks its respondents about their reading behavior regarding the abovementioned media. For each outlet, the AWA provides the respondent’s reading
frequency on an ordinal scale that ranges from 0 (“never/unknown”) to 5 (“all
issues”).
The media content data can be used to compute the share of opinionated
reports in the news coverage. I calculate this share separately for each medium, as
well as for the time periods corresponding to the distinct AWA waves. Reports are
treated as opinionated, if they contain an explicit or implicit judgment. According
to Media Tenor International, an explicit judgment refers to a clear evaluation of
the actor(s) or issue(s) by the means of positive/favorable/approving or
negative/unfavorable/disapproving words. Giving an implicit opinion is defined as
embedding the actor(s) or issue(s) in a judgmental context, for example, when a
report describes government spending as vote catching. As discussed in the theory
11
section, only reports with unambiguous, one-sided judgments are considered as
opinionated news.
[Table 1 here]
Table 1 shows the resulting shares. Accordingly, news outlets vary in their
provision of opinionated news in politics and economics. For instance, the
intellectual Die Zeit had an average share of 43.9% of opinionated reports in the
periods under consideration, whereas the tabloid Bild exhibited an average of
59.2%. Moreover, the shares of judgmental reports varied over time. While the
German print media provided opinionated articles in 57.6% of the cases in 2002,
two years later this share amounted to 47%.
Next, the shares of opinionated news are weighted by the respondents’
reading frequencies to obtain an indicator of the personal consumption of
opinionated news. If
medium at time , and
denotes the reading frequency of the individual for the
the share of opinionated news for the same medium
and time, the indicator can be written as:
∑
=
(1)
∑
The denominator accounts for individual-specific differences in the overall level
of news consumption, so that personal shares of opinionated news are comparable
across individuals, irrespective of general reading patterns. By construction,
may vary between 0 and 100%. The indicator only captures people who at least
sometimes consume print news. After subtracting the 15,465 nonreaders, the
sample comprises 70,292 respondents.
indicates that these respondents
consume one-sidedly opinionated news in 53.4% of the cases.
12
3.3 Explanatory Variables
The empirical goal of this study is to verify whether variation in the consumption
indicator can be explained by individual information-processing predispositions.
Due to the broad range of questions in the AWA surveys, several variables are
readily available to measure or approximate these predispositions:
•
A dummy variable that takes the value 1 for men and 0 for women serves
to verify hypothesis 1.
•
Verification of the second hypothesis is straightforward too. The AWA
provides an age variable that categorizes the respondents into 14 equally
divided age groups.
•
It is not possible to obtain a direct measure of cognitive ability from
survey data. However, the empirical literature on the need for cognition,
as well as on political sophistication, suggests various proxy variables;
for reviews, see Cacioppo et al. (1996) and Luskin (1987), respectively.
Based on these suggestions, I use dummy variable sets that indicate the
respondents’ school degree, kind of job training, and occupation group.
Hypothesis 3 is further evaluated by the respondents’ interest in politics
and economics (both measured on an integer scale ranging from 1 to 3),
and the households’ net income (in euro, measured on a scale with 13
equally divided categories).
•
A dummy variable set that captures the respondents’ religious
denomination serves to verify hypothesis 4a and 4b.
•
Hypothesis 5 is evaluated in two ways. First, a dummy variable indicates
whether a respondent resides in East or West Germany at the time of the
interview. Because this dummy variable fails to account for relocations
after reunification, another dummy variable set captures the respondents’
location in the fall of 1989.
13
4. Estimation and Results
4.1 Baseline Specification
The AWA is a series of independent samples of respondents. Therefore, it is
possible to estimate a model of repeated cross sections by ordinary least squares:
=
The dependent variable
+
+
(2)
+
denotes the indicator of the consumption of
opinionated news, which varies across individuals and over time . The vector
contains the explanatory variables described in the previous section. The
model also contains a vector of time dummies (
) to account for exogenous
variation in the amounts of opinionated news, such as related to general trends,
business cycles, or political constellations. Because
is based on equilibria of
supply and demand, it is not necessary to control for outlet- or issue-specific
characteristics (e.g., copy price or advertising expenditures). The ’s denote the
(vectors of) parameters to be estimated. The error term
is assumed to be
independently and normally distributed. However, the assumption of
homoscedastic residuals is relaxed; robust standard errors account for
heteroscedasticity that might be introduced by estimating equation (2) with the
sampling weights provided by the AWA.
[Table 2 here]
Table 2 shows the estimation results. Accordingly, men have significantly
higher levels of consumption of opinionated news than women, which confirms
hypothesis 1.
The coefficient of the age variable has the expected positive sign. It is not
significantly different from 0 though, so that hypothesis 2 cannot be confirmed.
The respondents’ cognitive abilities are approximated by a multitude of
variables. To begin with, people with a school degree other than higher secondary
14
education consume significantly more opinionated news. The decreasing
magnitude of the coefficients corresponds to the rank order of the degrees, being
largest for people with no degree (2.532) and smallest for respondents with a
technical one (0.412). The estimates regarding the respondents’ job training reveal
a similar pattern. People with no or vocational training consume more judgmental
reports than individuals with a university degree. Respondents with technical
training, however, exhibit lower consumption levels than university graduates.
Grouping people into occupation groups shows that farmers and blue-collar
workers exceed white-collar workers in the consumption of opinionated news;
other differences are not significant. Moreover, the consumption of judgmental
reports significantly decreases with rising interest in economics and especially
politics. Finally, the decline that is associated with increasing household income is
significant too. Hence the estimates confirm hypothesis 3, assuming that there is a
positive relationship between cognitive abilities on the one hand, and schooling,
vocational training, white-collar occupations, interest in politics/economics, and
income on the other.
The findings also confirm hypothesis 4b. Catholics consume more
opinionated news than Protestants, although the coefficient is significant at the 5%
level only. Other denominations do not differ from a Protestant one. People
without a religious affiliation exhibit significantly lower consumption levels than
Protestants – and therefore lower levels than people with a religious affiliation in
general – so that hypothesis 4a can be confirmed too.
Finally, the data support hypothesis 5. Having lived in West Germany in
the fall of 1989 implies a significantly higher consumption of opinionated news
than having lived in East Germany at this time. In addition to this socialization
effect, there is geographical divide, because the difference between people
currently living in East and West Germany is also significant.
15
4.2 Sensitivity Analysis
These results rest on crucial assumptions about the measurement of the individual
consumption of opinionated news. It is therefore advisable to evaluate
modifications in the dependent variable. The first modification involves weighting
daily and weekly media differently. In the baseline specification, the dependent
variable does not account for the frequency of publication; weekly media are
assumed to have the same contribution to the individuals’ share of opinionated
news as daily media. Although this is a reasonable assumption – because people
often read weekly magazines and newspapers over several days – a test can help
to verify it. For that purpose, an alternative version of the dependent variable
weights the contribution of the daily media in the sample six times higher than
that of the weekly outlets (i.e., to account for publication from Monday to
Saturday). The resulting variable is very similar to the baseline version; the
bivariate correlation amounts to 0.932 (p < .01). As the first column in Table 3
shows, the coefficient estimates are generally robust towards the different
weighting scheme. An exception is the age variable, which has a significantly
negative coefficient now. Moreover, the difference between Catholics and
Protestants is not significant anymore.
[Table 3 here]
The second modification of the dependent variable addresses the
assumption that ambivalent or two-sided news do not have the same heuristic
value as clear, one-sided reports. To check the validity of this assumption, an
alternative version of the dependent variable is constructed that uses all reports
that offer at least two different viewpoints to determine the shares of opinionated
news. According to this variable, people consume reports with multiple opinions
in about 16.4% of the cases. The alternative and baseline dependent variable are
16
negatively correlated (-0.133, p < .01), which suggests that people tend to prefer
either one-sided opinions or multiple viewpoints. The corresponding estimates are
almost a mirror image of those of the baseline specification, as the second column
in Table 3 indicates: Women, Protestants, people without a religious affiliation,
and West Germans consume more news with multiple viewpoints than their
counterparts; and the consumption of this kind of news increases with cognitive
abilities. Thus it can be concluded that processing news with one-sided opinions
requires less cognitive effort than processing reports that provide diverse views.
Third, the results depend on the accuracy of the content analyses
conducted by Media Tenor International. Although the identification of opinions
is subject to strict codebook criteria, as well as frequent tests of inter-coder
reliability and validity, the possibility of inaccuracies or bias caused by subjective
assessments of the coders cannot be ruled out completely. A more objective
alternative to the baseline version of the dependent variable can be constructed by
using counts of editorials and other explicit opinion articles. Based on this
identification procedure, the respondents consume opinionated news in about 10%
of the cases. The low bivariate correlation of this variable with the baseline
version (0.176, p < .01) suggests large differences in measuring openly declared
versus undeclared opinions. As the third column in Table 3 indicates, the estimates
remain fairly stable though. Only the coefficients for male, older, Catholic, and
West German respondents reveal major deviations from the baseline specification.
Considering the very low model fit (R2 = 0.178), however, it is likely that these
deviations result from the inaccurate way of capturing the amounts of opinionated
reports in the news.
Fourth, sampling weights are necessary to draw valid conclusions from the
data at hand. Because sampling weights may cause inefficient estimates due to
heteroscedasticity, it is important to calculate robust standard errors. These
standard errors tend to be larger than conventional ones though, which might lead
17
to false conclusions about the coefficients’ statistical significance. For this reason,
the final modification involves estimating the baseline specification without
sampling weights, in which case the calculation of conventional standard errors is
possible. As the fourth column in Table 3 shows, this alternative estimation
method provides results that are very similar to the baseline specification; the
coefficients do not differ substantially.
5. Conclusion
Overall then, there is no evidence for the expected effect of the respondents’ age
on the consumption of opinionated news (H2). The evidence for the hypothesized
difference between Catholics and Protestants (H4b) is weak. The baseline
specification provides the expected coefficient estimate, whereas most of the
alternative models do not. Considering the other hypotheses, the findings are
fairly robust though. The estimates confirm the presumed influence of the
information-processing predispositions regarding sex and gender (H1), cognitive
abilities (H3), religious affiliation (H4a), and east-west differences (H5).
Thus it can be concluded that opinionated reports with one-sided (but not
ambivalent) messages are of additional heuristic value to news consumers.
Individual predispositions matter; they determine whether unambiguous opinions
in the news coverage are valuable or disturbing to the recipient. Predispositions
towards systematic information processing increase the consumption of objective
or balanced news, whereas predispositions towards heuristic strategies favor the
consumption of news with one-sided opinions. In large parts, this relationship can
be interpreted causally, because people’s predispositions are based on
characteristics that are usually fixed (sex, gender, location of socialization), or
only change seldom or gradually (religion, cognitive abilities). In this
constellation it is very unlikely for media companies to influence consumer
preferences. Instead, individual predispositions determine the observed patterns in
18
the consumption of news. When it comes to the share of opinionated reports, the
demand side thus drives the market: Consumers choose their preferred kind of
news; media companies adjust their coverage to the preferences of the readers; or
both. Hence the findings support news-market theories of consumer-based profit
maximization (Mullainathan & Shleifer, 2005; Gentzkow & Shapiro, 2006, 2010;
Chan & Suen, 2008). Media companies determine how challenging the
consumption of their product is. Similar to many other product decisions – such as
choices over cover stories, headlines, illustrations, or free supplements – changing
the amount (more or less) and kind (one-sided or ambivalent) of opinion thus
serves as a sales instrument.
In contrast to previous, often experimental research on the use of
heuristics, this study’s evidence is based on large-scale observational data.
Although such a comprehensive approach has the advantage of a high level of
external validity, there are some limitations. In particular, the data only refer to
German, national print news about politics and economics. Further research is
necessary to evaluate whether the relationship between audience predispositions
and the opinionation of news holds in countries with other journalistic models, for
other types of media, and other topics.
19
References
Alesina, A., & Fuchs-Schündeln, N. (2007). Good-bye Lenin (or not?): The effect
of communism on people’s preferences. American Economic Review,
97(4), 1507–1528.
Andrew, B. C. (2007). Media-generated shortcuts: Do newspaper headlines
present another roadblock for low-information rationality? Press/Politics,
12(2), 24–43.
Barnhurst, K. G. (2003). The makers of meaning: National Public Radio and the
new long journalism, 1980–2000. Political Communication, 20(1), 1–22.
Barnhurst, K. G., & Mutz, D. (1997). American journalism and the decline in
event-centered reporting. Journal of Communication, 47(4), 27–52.
Bauernschuster, S., Falck, O., Gold, R., & Heblich, S. (2012). The shadows of the
socialist past: Lack of self-reliance hinders entrepreneurship. European
Journal of Political Economy, 28(4), 485–497.
Benson, R., & Hallin, D. C. (2007). How states, markets and globalization shape
the news: The French and US national press, 1965–97. European Journal
of Communication, 22(1): 27–48.
Boudreau, C., & McCubbins, M. D. (2010). The blind leading the blind: Who gets
polling information and does it improve decisions? Journal of Politics,
72(2), 513–527.
Boukes, M., Boomgaarden, H. G., Moorman, M., & De Vreese, C. H. (2014).
News with an attitude: Assessing the mechanisms underlying the effects of
opinionated news. Mass Communication and Society, 17(3), 354–378.
Buss, D. M. (1995). Psychological sex differences: Origins through sexual
selection. American Psychologist, 50(3), 164–168.
Cacioppo, J. T., Petty. R. E. (1982). The need for cognition. Journal of Personality
and Social Psychology, 42(1), 116–131.
20
Cacioppo, J. T., Petty, R. E., Feinstein, J. A., & Jarvis, W. B. G. (1996).
Dispositional differences in cognitive motivation: The life and times of
individuals varying in need for cognition. Psychological Bulletin, 119(2),
197–253.
Chaiken, S. (1980). Heuristic versus systematic information processing and the
use of source versus message cues in persuasion. Journal of Personality
and Social Psychology, 39(5), 752–766.
Chaiken, S., & Trope, Y. (Eds.). (1999). Dual-process theories in social
psychology. New York: Guilford Press.
Chan, J., & Suen, W. (2008). A spatial theory of news consumption and electoral
competition. Review of Economic Studies, 75(3), 699–728.
Choi, Y. (2010). Religion, religiosity, and South Korean consumer switching
behaviors. Journal of Consumer Behavior, 9(3), 157–171.
Craik, F. I. M., & Byrd, M. (1982). Aging and cognitive deficits: The role of
attentional resources. In F. I. M. Craik & S. Trehub (Eds.), Aging and
cognitive processes (pp. 191–211). New York: Plenum Press.
Dalton, R. J., Beck, P. A., & Huckfeldt, R. (1998). Partisan cues and the media:
Information flows in the 1992 presidential election. American Political
Science Review, 92(1), 111–126.
Darley, W. K., & Smith, R. E. (1995). Gender differences in information
processing strategies: An empirical test of the selectivity model in
advertising response. Journal of Advertising, 24(1), 41–56.
Delener, N. (1990). The effects of religious factors on perceived risk in durable
goods purchase decisions. Journal of Consumer Marketing, 7(3), 27–38.
Delener, N. (1994). Religious contrasts in consumer decision behaviour patterns:
Their dimensions and marketing implications. European Journal of
Marketing, 28(5), 36–53.
21
DellaVigna, S., & Gentzkow, M. (2010). Persuasion: Empirical evidence. Annual
Review of Economics, 2(1), 643–669.
Esser, F., & Umbricht, A. (2013). Competing models of journalism? Political
affairs coverage in US, British, German, Swiss, French and Italian
newspapers. Journalism, 14(8), 989–1007.
Feldman, L. (2011a). The opinion factor: The effects of opinionated news on
information processing and attitude change. Political Communication,
28(2), 163–181.
Feldman, L. (2011b). The effects of journalist opinionation on learning from the
news. Journal of Communication, 61(6): 1183–1201.
Fink, K., & Schudson, M. (2014). The rise of contextual journalism, 1950s–
2000s. Journalism, 15(1), 3–20.
Gentzkow, M., & Shapiro, J. M. (2006). Media bias and reputation. Journal of
Political Economy, 114(2), 280–316.
Gentzkow, M., & Shapiro, J. M. (2010). What drives media slant? Evidence from
U.S. daily newspapers. Econometrica, 78(1), 35–71.
Graber, D. A. (2001). Adapting political news to the needs of twenty-first century
Americans. In W. L. Bennett & R. M. Entman (Eds.), Mediated politics:
Communication in the future of democracy (pp. 433–452). Cambridge:
Cambridge University Press.
Kam, C. D. (2005). Who toes the party line? Cues, values, and individual
differences. Political Behavior, 27(2), 163–182.
Kuhn, A. (2012). Inequality perceptions, distributional norms, and redistributive
preferences in East and West Germany. German Economic Review, 14(4),
483–499.
Laroche, M., Saad, G., Cleveland, M., & Browne, E. (2000). Gender differences
in information search strategies for a Christmas gift. Journal of Consumer
Marketing, 17(6), 500–524.
22
Lau, R. R., & Redlawsk, D. P. (2001). Advantages and disadvantages of cognitive
heuristics in political decision making. American Journal of Political
Science, 45(4), 951–971.
Lau, S. (1989). Religious schema and values. International Journal of
Psychology, 24(2), 137–156.
Luskin, R. C. (1987). Measuring political sophistication. Journal of Political
Science, 31(4), 856–899.
McDaniel, S. W., & Burnett, J. J. (1990). Consumer religiosity and retail store
evaluative criteria. Journal of the Academy of Marketing Science, 18(2),
101–112.
Meyers-Levy, J. (1988). The influence of sex roles on judgment. Journal of
Consumer Research, 14(4), 522–530.
Meyers-Levy, J., & Loken, B. (2015). Revisiting gender differences: What we
know and what lies ahead. Journal of Consumer Psychology, 25(1), 129–
149.
Meyers-Levy, J., & Maheswaran, D. (1991). Exploring differences in males’ and
females’ processing strategies. Journal of Consumer Research, 18(1), 63–
70.
Meyers-Levy, J., & Sternthal, B. (1991). Gender differences in the use of message
cues and judgments. Journal of Marketing Research, 28(1), 84–96.
Mullainathan, S., & Shleifer, A. (2005). The market for news. American
Economic Review, 95(4), 1031–1053.
Mummert, A., & Schneider, F. (2001). The German shadow economy: Parted in a
united Germany. Public Finance Analysis, 58(3), 286–316.
Patterson, T., & Seib, P. (2005). Informing the public. In G. Overholser & K. Hall
Jamieson (Eds.), The press (pp. 189–202). Oxford: Oxford University
Press.
23
Putrevu, S. (2001). Exploring the origins and information processing differences
between men and women: Implications for advertisers. Academy of
Marketing Science Review, 10(1), 1–14.
Prior, M. (2013). Media and political polarization. Annual Review of Political
Science, 16(1), 101–127.
Reder, L. M., Wible, C., & Martin, J. (1986). Differential memory changes with
age: Exact retrieval versus plausible inference. Journal of Experimental
Psychology, 12(1), 72–81.
Rhee, J. W., & Capella, J. N. (1997). The role of political sophistication in
learning from news: Measuring schema development. Communication
Research, 24(3), 197–233.
Riggle, E. D. B., & Johnson, M. M. S. (1996). Age differences in political
decision making: Strategies for evaluating political candidates. Political
Behavior, 18(1), 99–118.
Schudson, M. (1978). Discovering the news: A social history of American
newspapers. New York: Basic Books.
Smith, G., & Searles, K. (2013). Fair and balanced news or a difference of
opinion? Why opinion shows matter for media effects. Political Research
Quarterly, 66(3), 671–684.
Torgler, B. (2003). Does culture matter? Tax morale in an East-West-German
comparison. Public Finance Analysis, 59(4), 504–528.
Wilke, J., & Reinemann, C. (2001). Do the candidates matter? Long-term trends
of campaign coverage: A study of the German press since 1949. European
Journal of Communication, 16(3), 291–314.
Yoon, C. (1997). Age differences in consumers’ processing strategies: An
investigation of moderating influences. Journal of Consumer Research,
24(3), 329–342.
24
Yoon, C., Cole, C. A., & Lee, M. P. (2009). Consumer decision making and aging:
Current knowledge and future directions. Journal of Consumer
Psychology, 19(1), 2–16.
Zaller, J. R. (1992). The nature and origins of mass opinion. Cambridge:
Cambridge University Press.
25
Table 1: Shares of Opinionated News across Outlets and over Time (%)
2002
2003
2004
2005
2006
2007
Frankfurter Allgemeine
53.1
46.4
43.9
50.4
56.0
55.2
50.8
Zeitung
Süddeutsche
Zeitung
56.4
47.5
45.7
52.2
55.6
55.7
52.2
Die Welt
57.9
55.6
48.1
56.8
58.5
60.4
56.2
Frankfurter Rundschau
59.1
49.0
47.2
52.4
53.3
58.0
53.2
Bild
60.5
54.1
49.9
65.7
57.4
67.7
59.2
Die Zeit
46.9
39.9
44.9
44.7
36.0
50.9
43.9
Spiegel
59.0
47.2
53.2
51.5
53.1
46.7
51.8
Stern
58.5
52.8
41.0
50.9
40.8
49.6
48.9
Focus
67.0
45.8
49.3
53.1
59.4
36.2
51.8
57.6
48.7
47.0
53.1
52.2
53.4
Notes: Own calculations based on data from Media Tenor International. Italic figures are row and
column means.
26
Table 2: Estimation Results
Coefficient
Standard error
Hypothesis
Male
0.668 ***
0.042
1
Age
0.008
0.007
2
School degree (ref.: higher secondary)
None
2.532 ***
0.149
3
Lower secondary
2.034 ***
0.065
3
Intermediate secondary
1.093 ***
0.060
3
Technical
0.412 ***
0.072
3
Job training (ref.: university degree)
No degree
0.718 ***
0.084
3
Vocational training
0.253 ***
0.065
3
Technical training
-0.229 ***
0.075
3
Occupation group (ref.: white collar)
Blue collar
0.888 ***
0.055
3
0.072
3
Civil servant
-0.027
Farmer
0.345 **
0.154
3
Self-employed
0.099
0.096
3
0.098
3
N/A
0.021
Interest in politics
-0.636 ***
0.033
3
Interest in economics
-0.386 ***
0.032
3
Household income
-0.029 ***
0.007
3
Denomination (ref.: Protestant)
Catholic
0.099 **
0.047
4
0.159
4
Other
0.014
None
-0.157 ***
0.050
4
Current residence West
-0.614 ***
0.111
5
Residence fall 1989 (ref.: East)
West
-0.719 ***
0.107
5
0.200
5
Abroad
0.110
Born after
0.409
0.382
5
No response
-0.097
0.233
5
Notes: The model contains an intercept and time dummies (output omitted). R2 = 0.490; N =
70,292.
*p < .10. **p < .05. ***p < 0.01.
27
Table 3: Alternative Estimates
1
0.845 ***
-0.016 **
Modification
2
3
-0.441 ***
-0.405 ***
0.005
0.011 **
4
Male
0.669 ***
Age
-0.004
School degree (ref.: higher secondary)
No degree
2.344 ***
-2.202 ***
0.967 ***
2.487 ***
Lower secondary degree
2.111 ***
-1.813 ***
0.944 ***
2.058 ***
-1.026 ***
0.532 ***
1.082 ***
Intermediate secondary degree
1.182 ***
Technical degree
0.462 ***
-0.526 ***
0.238 ***
0.421 ***
Job training (ref.: university degree)
-0.757 ***
0.340 ***
0.796 ***
No degree
0.773 ***
Vocational training
0.422 ***
-0.485 ***
0.104 **
0.361 ***
Technical training
-0.044
-0.088 *
-0.237 ***
-0.167 **
Occupation group (ref.: white collar)
Blue collar
0.809 ***
-0.713 ***
0.372 ***
0.893 ***
Civil servant
0.026
0.023
-0.090 *
0.099
Farmer
0.213
-0.295 ***
-0.147
0.267 **
-0.108 *
-0.014
0.125
Self-employed
0.134
N/A
-0.164 *
-0.098
0.152 **
-0.090
Interest in politics
-0.512 ***
0.508 ***
-0.364 ***
-0.631 ***
Interest in economics
-0.348 ***
0.340 ***
-0.254 ***
-0.337 ***
Household income
-0.026 ***
0.031 ***
-0.021 ***
-0.025 ***
Denomination (ref.: Protestant)
Catholic
-0.016
-0.028
0.025
0.111 ***
Other
-0.085
0.115
-0.086
-0.047
0.233 ***
-0.058 *
-0.148 ***
None
-0.202 ***
Current residence West
-0.523 ***
0.473 ***
0.412 ***
-0.629 ***
Residence fall 1989 (ref.: East)
0.793 ***
-0.022
-0.729 ***
West
-0.582 ***
Abroad
0.163
0.415 ***
0.135
0.055
Born after
0.137
0.556 ***
0.194
-0.122
0.488 **
0.460 ***
0.636 ***
No response
-0.191
R2
0.496
0.600
0.178
0.492
Notes: Modification 1: different weighting of daily and weekly media; modification 2: only
treating reports with multiple viewpoints as opinionated news; modification 3: shares of explicit
opinion articles instead of all opinionated reports; modification 4: calculation of conventional
standard errors and no sampling weights. All models contain an intercept and time dummies
(output omitted). N = 70,292.
*p < .10. **p < .05. ***p < 0.01.
28