737900

research-article20172017
RAP0010.1177/2053168017737900Research & PoliticsHopkins et al.

Research Article
Research and Politics

Does newspaper coverage influence or October-December 2017: 1­–7

© The Author(s) 2017
Reprints and permissions:
reflect public perceptions of the economy? sagepub.co.uk/journalsPermissions.nav
https://doi.org/10.1177/2053168017737900
DOI: 10.1177/2053168017737900
journals.sagepub.com/home/rap

Daniel J. Hopkins1, Eunji Kim2, and Soojong Kim3

Abstract
Citizens’ economic perceptions can shape their political and economic behavior, making the origins of those perceptions
an important question. Research commonly posits that media coverage is a central source. Here, we test that prospect
while considering the alternative hypothesis that media coverage instead echoes public perceptions. This paper applies
a straightforward automated measure of the tone of economic coverage to 490,039 articles from 24 national and local
media outlets over more than three decades. By matching the 245,947 survey respondents in the Survey of Consumer
Attitudes and Behavior to measures of contemporaneous media coverage, we can assess the sequencing of changes in
media coverage and public perceptions. Together, these data illustrate that newspaper coverage does not systematically
precede public perceptions of the economy, a finding which analyses of television transcripts reinforce. Neither national
nor local newspapers appear to strongly influence economic perceptions.

Keywords
Automated content analysis, economic perceptions, news media coverage, public opinion

Introduction
Americans’ perceptions of the economy are powerful
influences on their political and economic behavior (e.g.,
Hetherington, 1996; Gerber and Huber, 2009; Vavreck,
2009). At the same time, prior research indicates that the
impact of local economic conditions on perceptions, while
real, is limited in magnitude (Reeves and Gimpel, 2012;
Ansolabehere et al., 2014). That raises the question of
how Americans come to see the economy as strong or
weak (e.g., De Boef and Kellstedt, 2004), especially given
that assessing economic performance in a country of mil-
lions involves knowing about the conditions of strangers
in faraway places. Understandably, prior scholarship
points to mass media as a central source of Americans’
economic perceptions (e.g., Mutz, 1994; Blood and
Phillips, 1995; Soroka, 2006; Soroka et al., 2015).
Still, media influence on public perceptions is more
commonly assumed than demonstrated. While there is
extensive experimental evidence demonstrating the impact
of media coverage on various issue attitudes (e.g., Iyengar
and Kinder, 1987; Nelson et al., 1997), it is not clear that
more skeptical of media and elite rhetorical influence (e.g.,
Druckman and Nelson, 2003; Enns, 2014). Public inatten-
tion (Prior, 2007), selective information-seeking (Druckman
et al., 2012), and today’s increasingly fragmented media
market (Bennett and Iyengar, 2008) might all limit the
influence of specific media outlets in shaping public
perceptions.
A handful of existing studies also discuss one underex-
plored possibility: that public opinion influences media
coverage (Stevenson et al., 1994; Haller and Norpoth,
1997; Soroka et al., 2015). Haller and Norpoth (1997), for
instance, find that economic evaluations are similar for
respondents with and without significant news exposure,
suggesting that other factors shape economic perceptions.
In addition, given media outlets’ need to attract and
1Department of Political Science, University of Pennsylvania, USA
2Departmentof Political Science and Annenberg School for
Communication, University of Pennsylvania, USA
3Annenberg School for Communication, University of Pennsylvania, USA

such effects operate in real-world conditions (Gaines et al.,

Corresponding author:
2007; Barabas and Jerit, 2010). Some research finds that Daniel J. Hopkins, Department of Political Science, University of
media coverage can profoundly shape public attitudes Pennsylvania, 208 S. 37th Street, Philadelphia, PA 19104-6215, USA.
(Gilens, 1999; Kellstedt, 2003), while other research is Email: danhop@sas.upenn.edu

Creative Commons Non Commercial CC BY-NC: This article is distributed under the terms of the Creative Commons
Attribution-NonCommercial 4.0 License (http://www.creativecommons.org/licenses/by-nc/4.0/) which permits non-commercial
use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and
Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage).
2 Research and Politics 
maintain an audience, it is plausible that their coverage will
follow public opinion, telling viewers and readers what
they already believe to expand and maintain market share
(George and Waldfogel, 2006; Gentzkow and Shapiro,
2010). In this approach, media coverage still varies in tone,
but public opinion chiefly shapes media coverage rather
than the reverse.
Here, we assess these two explanations of the relation-
ship between economic coverage and economic percep-
tions. Building on prior research on the agenda-setting
role of print journalism (e.g., Roberts and McCombs,
1994), we first examine newspaper stories mentioning the
economy in the New York Times between 1980 and 2015
and the Washington Post between 1978 and 2015. This
amounts to 84,071 stories in the Post and 124,285 stories
in the Times. From these stories, the paper develops a sim-
ple, dictionary-based metric of economic coverage that
proves valid and reliable. In fact, an independent assess-
ment by Barberá et al. (2016) finds that this metric is more
highly correlated with key economic indicators than the
dictionary-based metrics of Thelwall et al. (2010) and
Soroka et al. (2015). We also apply the same metric to
269,910 transcripts/articles from 2 national television net-
works (ABC and CBS), 19 state-level newspapers, and
USA Today. To measure public economic perceptions, we
use the Surveys of Consumer Attitudes and Behavior
(SCA), a monthly survey which provides 245,947
respondents between 1978 and 2015.1 In short, this paper
builds on prior research in at least three ways: it presents
a new measure of the tone of economic coverage which is
at once valid and easily estimated; it analyzes a broader
range of media outlets than prior work; and it updates
prior research by analyzing data through 2015.
The results indicate that coverage of the economy in
national newspapers is not a strong influence on subse-
quent public perceptions of economic performance.
Instead, Granger tests show that public economic evalua-
tions typically lead media coverage. We also contribute to
the existing literature by testing our hypothesis at the state
level using local media outlets. These results are less
definitive, but suggest that state-level newspapers may be
quicker to respond to changes in economic conditions than
national media outlets.
Together, these analyses provide evidence about what
is not influencing economic perceptions, but do not offer
convincing evidence about what does shape perceptions.
Americans’ economic perceptions do vary systematically
with economic conditions. We see, too, that these percep-
tions shift prior to national newspaper coverage. One
explanation is that Americans are able to extract eco-
nomic information from media coverage regardless of its
tone; a second is that other sources of information—
including media outlets with narrower, more targeted
audiences, as well as actual economic conditions or social
networks—prove influential.
Data and Methods
To explore the connection between media coverage and public
perceptions of the national economy, we combine two data
sources. Prior research indicates that television coverage com-
monly follows coverage in the major national newspapers
(e.g., Roberts and McCombs, 1994; Blood and Phillips, 1995),
making those newspapers “most likely” cases for influence.
Accordingly, we first measure the tone of coverage in two
national newspapers, the Times and Post, as research has
repeatedly found that these outlets are often agenda setters
(McCombs and Shaw, 1972; McCombs, 2005). Additionally,
these two have among the longest Lexis-Nexis archives.
One limitation in past research has come from the diffi-
culty of hand-coding documents on a large scale. The pro-
digious coding efforts of DeBoef and Kellstedt (2004) and
Soroka (2006) categorize under 5000 articles. A small but
growing number of recent studies utilize automated content
analyses that substantially reduce the costs of analyzing
large text data sets. Soroka et al. (2015) were able to ana-
lyze about 30,000 economic news stories in the US, and
Kayser and Peress (2015) collected over 2 million articles
in 16 different countries (see also Young and Soroka, 2012).
Here, we also use automated content analysis to include
any article in the Times or Post that used a series of words
related to economic performance.2 The advantage of using
coverage from throughout the newspaper is that Americans’
economic information is not limited to what they read in
explicitly political articles: business sections might provide
a wealth of economic information.
We extracted the 111,805 relevant Times articles and
48,052 relevant Post articles from any section in either news-
paper.3 The analysis then reduced each word to its common
stem and calculated the share of articles using each word stem
for each month. Experimentation showed a simple additive
index of 21 economic words.4 This procedure produces meas-
ures of economic coverage that prove both valid and reliable
(for a similar metric, see Kayser and Peress, 2015).5 In par-
ticular, Barberá et al. (2016) report that a very similar metric
is more strongly correlated with aggregate economic condi-
tions than other dictionary-based metrics—and that our meas-
ure’s performance rivals that of more sophisticated statistical
models. The resulting Times index varies from 0.108 to 2.332
while the Post index runs from 0.306 to 3.284 (see Appendix
A, Table 4). Higher values indicate a more negative tone.
As most economic indicators are heavily trended, a
measure which indicates very positive coverage in one
month and very negative coverage in the next month is
likely to be inaccurate. The Post-based measure correlates
with itself in the prior month at 0.769 and the Times-based
measure does so at 0.829. As with the economy itself, eco-
nomic coverage as measured through word usage appears
consistent from month to month. These measures correlate
with other conventional measures of economic perfor-
mance as well, ranging from the national unemployment
Hopkins et al. 3
rate (0.43 for the Times and 0.47 for the Post) to GDP
growth (−0.54 and −0.47), consumer sentiment about the
economy (−0.53 and −0.48), and stock market performance
(−0.17 and −0.11). These measures of negative economic
coverage act as we would expect, taking their highest val-
ues in the early 1980s, the early 1990s, the period just after
11 September 2001, and the run-up to the 2008 Great
Recession. Their lowest values come at times of strong eco-
nomic performance, such as the late 1990s and mid-2000s.
To measure economic perceptions, we turn to the SCA.
The survey has asked the same battery of questions about per-
sonal and sociotropic economic conditions every month since
January 1978. We combine four questions about economic
conditions into an index such that higher values indicate neg-
ative economic assessments, with question wording and sum-
mary statistics located in Appendix B. The index of consumer
sentiment ranges from a minimum of 5 (no concern about the
economy in any of the questions) to a maximum of 25 (high
concern about the economy in all questions). We average the
responses by month, yielding a time-series with 432 separate
months. The average month has 569.3 survey respondents.6
Those wishing to see how public opinion varies when
presented with different issue frames can and do make use
of survey experiments. Researchers wishing to analyze the
possibility of a reverse relationship—an impact of public
attitudes on media coverage—have a more challenging
task, as there is no clear, randomization-based design to
use. This paper thus draws on relative timing (Granger,
1988) to establish ceilings for the likely influence of news-
papers’ frames on public perceptions and vice versa. To the
extent that a causal relationship exists between two factors
that vary over time, changes in the causal variable should
correspond to subsequent changes in the affected variable.
If the tests illustrate that one variable systematically pre-
cedes another, the causal relationship is plausible but not
proven: that same temporal pattern could result from an
omitted causal agent affecting both measures. As a conse-
quence, we test one precondition of a causal relationship
rather than testing the causal relationship itself.
Results: Leading or Lagging?
Figure 1 presents two key measures—media economic con-
cern and public economic concern—with panels corre-
sponding to the Times and Post, respectively. The overall
trends are very similar. There appears to be something
approaching a national consensus about the state of the
economy, one that pervades both national reporting and
public opinion. Still, in identifying how plausible causal
claims about media coverage and public perceptions are,
the ordering of the time-series—whether the peaks and
nadirs of public opinion follow those in media coverage—
is critical. Looking at Figure 1, we do not see obvious cases
where changes in media coverage preceded shifts in public
perception. On the contrary, media coverage lags changes
in public attitudes at key moments, such as in the early
1980s and 2005 to 2006. To test these relationships for-
mally, Table 1 presents Granger tests.7
Its top panel displays the results for tests of the possibility
that the temporal pattern from media economic concern to pub-
lic economic concern is the result of chance alone. The Times
shows occasionally significant effects when we use lags of two
and three months, but no such findings with a one-month lag, a
pattern which is at odds with claims of media influence given
prior research indicating that media effects typically decay
quickly (e.g., Hill et al., 2013). The fact that the Post shows no
significant pattern for one, two, or three-month lags reinforces
the claim that economic coverage in major, national newspa-
pers does not systematically lead public economic perceptions.
The causal ordering implied by many media effects studies—
that shifts in elite frames should precede public opinion shifts—
is not sustained with respect to economic attitudes.
But is there evidence of the reverse relationship? The bot-
tom panel of Table 1 presents the results of Granger tests con-
sidering the alternative possibility that public opinion leads
media coverage. This time, every one of the six tests is sig-
nificant ( p < .01), providing robust evidence that Americans’
views of the economy shift before the relevant newspaper
coverage does. A separate OLS model indicates that as public
economic concern rises by 1.00—nearly a standard devia-
tion—we should expect Times coverage to become 0.03 more
negative, or 8% of its standard deviation (see Appendix C,
Table 8). As Appendix C, Table 7 shows, this pattern also
holds when breaking out respondents by their education or
income levels, without consistent evidence that any group is
more responsive to media coverage or more likely to antici-
pate media coverage (see also Haller and Norpoth, 1997).8
As Table 2 shows, we find the same pattern when repli-
cating these analyses using television transcripts from ABC
and CBS News. In both cases, the Granger test is signifi-
cant when public perceptions lead media coverage by one
month, but not when they lag.9 To address the possibility
that the Times and Post might be influenced by their par-
ticular media markets, we also considered USA Today using
a corpus from 1989 to 2016 and found similar results. To be
sure, this evidence is not sufficient to claim that public atti-
tudes cause media coverage, but it is consistent with that
claim. It is also consistent with the possibility that public
opinion and media coverage are both responding to another
factor—perhaps real-world economic conditions.
Additionally, we tested the same hypothesis at the state
level using 241,270 articles from 19 different local newspa-
pers from 17 large US states.10 In those tests, we restrict our
sample of survey respondents to those living in a specific
state, meaning that our measurement of public opinion is
substantially noisier. Even so, Granger tests indicate a sta-
tistically significant relationship in which public sentiment
follows media coverage in 16 of 19 examples, while media
coverage lags behind public opinion in 12 of 19 examples,
as Table 2 below details. (See Appendix A, Tables 2 and 4
4 Research and Politics 

Figure 1.  Public and media economic concern over time.

Note: It presents the average tone of economic coverage for the two national newspapers (The New York Times from 1980 to 2015 for the top
panel and The Washington Post from 1978 to 2015 for the bottom panel) as well as public economic assessments. The y-axis on the left reflects public
economic concern (the survey-based measure), while the y-axis on the right reflects media economic concern.

for descriptions of the data.) In sum, the results at the state
level are less conclusive, and suggest that perhaps state-
level newspapers play more of a role in shaping public
economic perceptions. Given that newspaper organizations
are strategic actors which respond to both professional
norms and economic incentives (Hamilton, 2004; George
and Waldfogel, 2006), it is plausible that locally targeted
newspapers may have more incentive to emphasize local
economic changes.
Conclusion
Public perceptions about the economy do not appear to
originate with the tone of national newspaper coverage.
Hopkins et al. 5

Table 1.  Granger test results (two national newspapers).

Model Granger Test F-Statistic p-value

Lag 1 Media economic concern (Times) → Public economic concern 0.978 0.323
  Media economic concern (Post) → Public economic concern 0.000 0.999
Lag 2 Media economic concern (Times) → Public economic concern 3.570* 0.029
  Media economic concern (Post) → Public economic concern 2.090 0.125
Lag 3 Media economic concern (Times) → Public economic concern 3.223* 0.023
  Media economic concern (Post) → Public economic concern 2.308 0.076
Lag 1 Public economic concern → Media economic concern (Times) 9.365** 0.002
  Public economic concern → Media economic concern (Post) 17.771*** 0.000
Lag 2 Public economic concern → Media economic concern (Times) 3.321* 0.037
  Public economic concern → Media economic concern (Post) 6.754** 0.001
Lag 3 Public economic concern → Media economic concern (Times) 2.394 0.068
  Public economic concern → Media economic concern (Post) 4.864** 0.003

Note: For lags of 1, 2, and 3 months, this table shows the results of Granger tests considering the possibility that the tone of newspaper coverage
moves prior to public economic assessments and vice versa. The table presents the F-statistic and the p-value corresponding to the hypothesis that
there is no directional relationship.

Table 2.  Granger Test Results (Newspapers and National TV).

Media → Public Public → Media

Media Source State F-Statistic p-value F-Statistic p-value
ABC News N/A 0.668 0.414 5.754* 0.017
Arkansas Democrat-Gazette Arkansas 3.958* 0.048 0.088 0.767
Atlanta Journal-Constitution Georgia 11.500*** 0.001 7.923** 0.006
Boston Globe Massachusetts 25.707*** 0.000 6.770* 0.010
CBS News N/A 2.998 0.085 28.991*** 0.000
Chattanooga Times Free Press Tennessee 0.003 0.960 1.436 0.234
Chicago Sun-Times Illinois 13.204*** 0.000 3.921* 0.050
Cleveland Plain Dealer Ohio 5.513* 0.020 9.976** 0.002
Columbus Dispatch Ohio 7.361** 0.008 11.220** 0.001
Herald Sun North Carolina 3.607 0.060 5.738* 0.018
Houston Chronicle Texas 6.177* 0.014 6.353* 0.013
Minnesota Star Tribune Minnesota 8.871** 0.004 12.819*** 0.001
New York Daily News New York 4.006* 0.048 16.165*** 0.000
Omaha World-Herald Nebraska 0.686 0.409 0.091 0.764
Oregonian Oregon 13.302*** 0.000 0.247 0.620
Pittsburgh Post-Gazette Pennsylvania 4.053* 0.046 0.028 0.867
San Diego Union Tribune California 30.586*** 0.000 32.075*** 0.000
San Francisco Chronicle California 5.907* 0.016 28.632*** 0.000
Seattle Post-Intelligencer Washington 9.420** 0.003 0.084 0.772
St. Petersburg Times Florida 20.522*** 0.000 25.995*** 0.000
USA Today N/A 1.726 0.190 7.744** 0.006
Virginia Pilot Virginia 6.894* 0.011 0.298 0.587

Note: This table displays the results of Granger tests with a one-month lag for various state-level media outlets, two national television networks,
and one national newspaper (USA Today). For these tests, we create separate public economic concern indices for each state which run to 2008. For
the national television networks, we use the resulting indices through 2010.

Perhaps the public is able to extract information about
economic performance irrespective of the tone of the
corresponding media coverage. The economy might be
an exception to the general pattern of low levels of
political information, as it has a substantial impact on
various aspects of Americans’ day-to-day lives. Indeed,
MacKuen et al. (1992) find that collectively, Americans
cast votes based on their economic expectations, a fact
which demands considerable economic knowledge on
the part of at least some individuals. On a related note,
Soroka et al. (2015) find that the public’s prospective
economic evaluations often precede shifts in the tone of
media coverage. As a result, this paper lends new weight
to scholarship contending that news outlets have a more
limited capacity to shape public perceptions than is often
appreciated. While a comprehensive account of public
6 Research and Politics 
perceptions of the economy would necessarily include
an assessment of responses to real-world economic con-
ditions (e.g., Hopkins, 2011), our goal here is simply to
assess the sequencing of public perceptions and media
accounts.11
There are methodological limitations worth acknowl-
edging. We have studied only a subset of media outlets,
with no attention to cable television, radio, social media,
newspapers with smaller circulations, or other would-be
sources of economic information. Estimated media cov-
erage sentiment varies with the corpus and keywords
employed (Barberá et al., 2016), so our measurement
strategy may be application-specific. But even with these
caveats, it is worth highlighting that our measure of pub-
lic economic concern associates strongly with real-world
economic statistics. In light of a somewhat surprising
finding that economic coverage by major national news-
papers has little effect on public perceptions, linking spe-
cific respondents to the media sources they use and
analyzing a wider range of media sources are important
next steps in advancing our understanding of the role of
media coverage in shaping economic perceptions. In
addition, as national newspapers tend to shape elite opin-
ion more directly than mass-level opinion (see Herbst,
1998), future research might further untangle the com-
plex interplay between media organizations, political
elites, and public economic perceptions. With respect to
methodology, these results underscore the value of
straightforward dictionary-based automated content anal-
ysis in specific applications.
Acknowledgements
The authors thank Tiger Brown, Owen O’Hare, Saleel Huprikar,
Thomas Munson, Gabrielle Rothschild, Anton Strezhnev, and
Elena Zhao for excellent research assistance. Yph Lelkes and
Diana Mutz kindly provided feedback.
Declaration of Conflicting Interest
The author(s) declared no potential conflicts of interest with respect
to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research,
authorship, and/or publication of this article.
Supplementary materials
The supplementary files are available at http://journals.sagepub.com/
doi/suppl/10.1177/2053168017737900. The replication files are
available at http://thedata.harvard.edu/dvn/dv/researchandpolitics.
Notes
  1. The length of this time-series helps ensure that the findings
are not driven by a specific presidency, business cycle, or
other events.
  2. The Lexis-Nexis search string was: “economy AND (boom
OR hire OR surplus OR employment OR profits OR debt OR
deficit OR downturn OR jobless OR recession OR slump OR
unemployment).”
 3. While Soroka et al. (2015) also analyze these two news-
papers, our searches were more permissive and so return a
markedly larger number of articles. Nonetheless, the monthly
correlation in the number of articles between their data and
ours is 0.68 (for the Times) and 0.46 (for the Post).
  4. These include 15 negative word stems (“bad,” “bear,” “debt,”
“drop,” “fall,” “fear,” “jobless,” “layoff,” “loss,” “plung,”
“problem,” “recess,” “slow,” “slump,” and “unemploy”) and
6 positive word stems (“bull,” “grow,” “growth,” “inflat,”
“invest,” and “profit”. See Appendix A, Table 3.). We intended
this index to place more weight on negative words to reflect
the media’s well-established tendency to over-report negative
economic news (Blood and Phillips, 1995; Broome, 2006)
as well as citizens’ tendency to be more sensitive to negative
information (Rozin and Royzman, 2001). We constructed the
media economic concern index by summing the frequencies of
negative and positive words and then subtracting them.
  5. We pursued other approaches, including the development of
a predictive model using 600 hand-coded articles and the use
of factor analysis to identify positive or negative economic
coverage. It is important to note, however, that measures
which are highly accurate at the level of the individual article
may nonetheless perform poorly in assessing our quantity of
interest, the aggregate tone of economic coverage (see also
Hopkins and King, 2010). The measure detailed here has the
twin virtues of validity and simplicity. Factor analyses gener-
ate similar results, as reported in Appendix A, Table 1.
  6. Hopkins (2011) demonstrates that the time-series are gener-
ally quite similar across income and educational groups, a
finding reinforced by the figures in Appendix D. Accordingly,
we do not use survey weights in constructing these measures.
 7. We have also run Granger tests using the first dimension
from factor analyses. See Appendix C, Tables 6 and 7.
 8. Additionally, we find that the pattern is consistent when
splitting the data before and after 1996, suggesting that these
results are not driven by the fragmentation of the media mar-
ket in recent decades.
  9. ABC News transcripts are available for 248 months; CBS
News transcripts are available for 227 months.
10. We employed large states to have sufficient sample sizes
from the Surveys of Consumer Attitudes and Behavior, and
then utilized the largest state-level newspapers archived in
Lexis-Nexis.
11. Tables 9 and 10 in Appendix C provide a preliminary assess-
ment along these lines, showing that both public economic
perceptions and media coverage appear to lead the unem-
ployment rate. What’s more, separate assessments indicate
that on average, coverage in the national outlets studied here
tends to lead the tone of state-level coverage. Given the
variability in state-level economic conditions, and given the
role of the Times and Post as outlets that endeavor to cover
national as well as regional and local affairs, such a finding
makes sense.
Carnegie Corporation of New York Grant
This publication was made possible (in part) by a grant from
Carnegie Corporation of New York. The statements made and
views expressed are solely the responsibility of the authors.
Hopkins et al. 7
References
Ansolabehere S, Meredith M and Snowberg E (2014) Mecro-
economic voting: Local information and micro-perceptions of
the macro-economy. Economics and Politics 26(3): 380–410.
Barabas J and Jerit J (2010) Are survey experiments externally
valid? American Political Science Review 104(2): 226–242.
Barberá P, Boydstun A, Linn S, McMahon R and Nagler J
(2016) Methodological challenges in estimating tone:
Application to news coverage of the U.S. economy.
Available at: https://pdfs.semanticscholar.org/8b52/9b50b
6bad5d3e3e59e1ab187c6de26dec86b.pdf.
Bennett WL and Iyengar S (2008) A new era of minimal effects?
The changing foundations of political communication.
Journal of Communication 58(4): 707–731.
Blood DJ and Phillips PCB (1995) Recession headline news, con-
sumer sentiment, the state of the economy and presidential
popularity: A time series analysis 1989–1993. International
Journal of Public Opinion Research 7(1): 2–22.
Broome A (2006) Setting the fiscal policy agenda: economic news
and election year tax debates in New Zealand. Law Context:
A Socio-Legal Journal 24(2): 60–77.
De Boef S and Kellstedt PM (2004) The political (and economic)
origins of consumer confidence. American Journal of
Political Science 48(4): 633–649.
Druckman JN, Fein J and Leeper TJ (2012) A source of bias in
public opinion stability. American Political Science Review
106(2): 430–454.
Druckman JN and Nelson KR (2003) Framing and deliberation:
How citizens’ conversations limit elite influence. American
Journal of Political Science 47(4): 729–745.
Enns PK (2014) The public’s increasing punitiveness and its influ-
ence on mass incarceration in the United States. American
Journal of Political Science 58(4): 857–872.
Gaines BJ, Kuklinski JH and Quirk PJ (2007) The logic of the sur-
vey experiment reexamined. Political Analysis 15(1): 1–20.
Gentzkow M and Shapiro JM (2010) What drives media slant?
Evidence from US daily newspapers. Econometrica 78(1):
35–71.
George LM and Waldfogel J (2006) The New York Times and
the market for local newspapers. The American Economic
Review 96(1): 435–447.
Gerber AS and Huber GA (2009) Partisanship and economic
behavior: Do partisan differences in economic forecasts
predict real economic behavior? American Political Science
Review 103(3): 407–426.
Gilens M (1999) Why Americans Hate Welfare: Race, Media, and
the Politics of Anti-Poverty Policy. Chicago, IL: University
of Chicago Press.
Granger CWJ (1988) Some recent developments in a concept of
causality. Journal of Econometrics 39(1–2): 199–211.
Haller HB and Norpoth H (1997) Reality bites: News exposure and
economic opinion. Public Opinion Quarterly 61(4): 555–575.
Hamilton J (2004) All the News That’s Fit to Sell: How the Market
Transforms Information into News. Princeton, NJ: Princeton
University Press.
Herbst S (1998) Reading Public Opinion: How Political Actors
View the Democratic Process. Chicago, IL: University of
Chicago Press.
Hetherington MJ (1996) The media’s role in forming vot-
ers’ national economic evaluations. American Journal of
Political Science 40(2): 372–395.
Hill SJ, Lo J, Vavreck L and Zaller J (2013) How quickly we
forget: The duration of persuasion effects from mass com-
munication. Political Communication 30(4): 521–547.
Hopkins DJ (2011) Whose economy? Perceptions of national eco-
nomic performance during unequal growth. Public Opinion
Quarterly 76(1): 50–71.
Hopkins DJ and King G (2010) A method of automated nonpara-
metric content analysis for social science. American Journal
of Political Science 54(1): 229–247.
Iyengar S and Kinder DR (1987) News that Matters. Chicago, IL:
University of Chicago Press.
Kayser MA and Peress M (2015) The media, the economy, and
the vote. Available at: https://www.sas.upenn.edu/polisci/
sites/www.sas.upenn.edu.polisci/files/PRAPC_Peress.pdf.
Kellstedt PM (2003) The Mass Media and the Dynamics
of American Racial Attitudes. New York: Cambridge
University Press.
MacKuen MB, Erikson RS and Stimson JA (1992) Peasants or
bankers? The American electorate and the U.S. economy.
American Political Science Review 86(3): 597–611.
McCombs ME (2005) A look at agenda-setting: Past, present and
future. Journalism Studies 6(4): 543–557.
McCombs ME and Shaw DL (1972) The agenda-setting function
of mass media. Public Opinion Quarterly 36(2): 176–187.
Mutz DC (1994) Contextualizing personal experience: The role of
mass media. Journal of Politics 56(3): 689–714.
Nelson TE, Clawson RA and Oxley ZM (1997) Media framing of
a civil liberties conflict and its effect on tolerance. American
Political Science Review 91(3): 567–583.
Prior M (2007) Post-Broadcast Democracy: How Media Choice
Increases Inequality in Political Involvement and Polarizes
Elections. New York, NY: Cambridge University Press.
Reeves A and Gimpel JG (2012) Ecologies of unease: Geographic
context and national economic evaluations. Political
Behavior 34(3): 507–534.
Roberts M and McCombs M (1994) Agenda setting and politi-
cal advertising: Origins of the news agenda. Political
Communication 11(3): 249–262.
Rozin P and Royzman EB (2001) Negativity bias, negativity dom-
inance, and contagion. Personality and Social Psychology
Review 5(4):296–320.
Soroka SN (2006) Good news and bad news: Asymmetric
responses to economic information. Journal of Politics
68(2): 372–385.
Soroka SN, Stecula DA and Wlezien C (2015) It’s (change in) the
(future) economy, stupid: Economic indicators, the media,
and public opinion. American Journal of Political Science
59(2): 457–474.
Stevenson RL, Gonzenbach WJ and David P (1994) Economic
recession and the news. Mass Communication Review 21(1):
4–19.
Thelwall M, Buckley K, Paltoglou G, Cai D and Kappas A
(2010) Sentiment strength detection in short informal text.
Journal of the American Society for Information Science and
Technology 61(12): 2544–2558.
Vavreck L (2009) The Message Matters: The Economy and
Presidential Campaigns. Princeton, NJ: Princeton University
Press.
Young L and Soroka SN (2012) Affective news: The auto-
mated coding of sentiment in political texts. Political
Communication 29(2): 205–231.