Violence Against Politicians, Negative Campaigning, and Public Opinion: Evidence From Poland

AuthorKrzysztof Krakowski,Juan S. Morales,Dani Sandu
DOIhttp://doi.org/10.1177/00104140211066211
Published date01 October 2022
Date01 October 2022
Subject MatterArticles
Article
Comparative Political Studies
2022, Vol. 55(12) 20862118
© The Author(s) 2022
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/00104140211066211
journals.sagepub.com/home/cps
Violence Against
Politicians, Negative
Campaigning, and Public
Opinion: Evidence From
Poland
Krzysztof Krakowski
1
, Juan S. Morales
2
and
Dani Sandu
3
Abstract
It is commonly viewedthat violence against politicians increases support for the
victims party. We revisit this conjecture drawing on evidence from an as-
sassination of an opposition politician in Poland. First, we analyze engagement
with Twitter content posted by opposition and government politicians using a
difference-in-differences framework. Second, we use a public opinion survey
collected in the days around the attack and compare party preferences of
respondentsinterviewed just before and respondents interviewedjust after the
attack. Our resultsreveal decreased support forthe victims (opposition) party
relative to support for the government. To explain this f‌inding, we show that
the oppositionantagonized the publicby engaging in negative campaigningagainst
the government over theirpoliticians assassination. Content analysis of tweets
and news media conf‌irms that citizens punished the opposition for their
negative campaigning after the violence. Tentative evidence suggests that these
effects could have had long-run political consequences.
Keywords
violence, public opinion, negative campaigning, Poland
1
Collegio Carlo Alberto, University of Turin, Torino, Italy
2
Wilfrid Laurier University, Waterloo, ON, Canada
3
European University Institute, Fiesole, Italy
Corresponding Author:
Krzysztof Krakowski, Collegio Carlo Alberto, University of Turin, Torino, Italy.
Email: krzysztof.krakowski@carloalberto.org
Introduction
Violence against politicians poses a serious threat to political order, even in the
most advanced liberal democracies. Appleton (2000) lists over 200 assas-
sination attempts against heads of government in the 20th century. Between
1950 and 2004, in two out of three years a national leader was assassinated
(Jones & Olken, 2009). Since 2016, high-level politicians from three EU-
member states have been victims of political violence: in 2016, a UK Labour
Party deputy, Jo Cox was shot to death few days before the Brexit vote; in
January 2019, PawełAdamowicz, the Mayor of Gda ´
nsk in Poland was
stabbed during a charity event; and in June 2019, Walter Lübcke, a regional
leader of the ruling German party was killed in front of his house. A recent
study shows that a surprising 13% of American mayors acknowledge having
been the victims of physical violence (Thomas et al., 2019). The frequency of
these incidents raises questions about the impact of violence against politi-
cians on public opinion.
At f‌irst glance, politiciansassassinations lead to an increase in support for
the targeted party. The assassinations of heads of states and governments have
been shown to reinforce incumbent support (Raviv et al., 1998;Appleton,
2000). These changes are usually explained by a higher government en-
dorsement in situations of threat and uncertainty (rally effect; Baker & Oneal,
2001), and a greater affect for victims of violence (empathy effect; Dinas et al.,
2016). In the case of incumbent assassinations, both mechanisms work in the
governments favor. However, when a nonincumbent politician is assassi-
nated, the two mechanisms work in opposite directions, leading to ambivalent
predictions regarding net changes in support.
Does approval for the victims party also increase in the case of nonin-
cumbent assassinations? We study this question drawing on evidence from a
lethal and tragic attack on the opposition mayor of Gda ´
nsk, PolandPaweł
Adamowiczwho was assassinated in January 2019 during a public charity
event. The killer explicitly mentioned Adamowiczs party aff‌iliation as a
reason for his assassination. We track changes in public opinion related to that
event in two ways. First, we analyze engagement with Twitter content posted
by opposition and government politicians using a difference-in-differences
framework. Second, we use a public opinion survey collected in the days
around the attack, in an Unexpected Event during Surveys Design (UESD)
framework, which compares respondents interviewed just before and those
interviewed just after the attack (see Muñoz et al., 2020).
Contrary to the common view in the literature, we f‌ind that the assassi-
nation of the Polish mayor decreased relative support for his opposition party.
Estimates are sizable: within a 15-day time window around the date of the
violent attack, government politicians received 35% higher Twitter engage-
ment compared to opposition politicians. In the survey data, within the 4-day
Krakowski et al. 2087
time window covered, we f‌ind that the government had a relative increase in
support of around 41%. We conf‌irm these f‌indings through a series of per-
mutation and falsif‌ication tests, including time and unit f‌ixed-effects models,
matching techniques, and placebo outcomes.
Why did the opposition lose support after the assassination of their own
politician? The stabbing of PawełAdamowicz marked an unprecedented
escalation of political conf‌lict in Poland. While citizens expected leaders to
attempt to de-escalate conf‌lict following the event, the opposition responded
through aggressive language and verbal attacks on the government (see Sarna
& Tyc, 2020). Consistent with this anecdotal evidence, we show that the loss
of support for the opposition is linked to its confrontational response to the
mayors assassination in the form of negative campaigning against the
government. We f‌ind that the oppositions use of negative campaigning after
the violence on Twitter explains about 60% of their loss of relative support.
This pattern is also present in the survey data. The oppositions loss of support
is most pronounced among survey respondents who watched partisan TV
channels and were thus most exposed to the partys negative campaigning.
Among neutral TV viewers, by contrast, the support for the opposition is even
slightly increased. Content analysis of 3221 news scraped from the main
Polish TV channels conf‌irms that these neutral outlets were the least likely to
broadcast the oppositions negative campaigning after the violence.
The relevance of the case study and the proposed mechanism are further
highlighted by tentative evidence on long-run patterns of political support. We
show that the 2019 electoral votes of opposition politicians and their negative
campaigning after the assassination are negatively correlated, suggesting that
shocks to party support following political assassinations could have long-run
political consequences. Importantly, these consequences are likely to depend
on how politicians respond to such tragic events.
Our study contributes to the literature in four important ways. First, we
provide evidence on the consequences of political assassinations in the case of
an attack on a nonincumbent politician. It is a common, yet understudied type
of violence against politicians, which may result in different changes in public
opinion than well-documented incumbent assassinations.
1
Second, we pro-
vide evidence on heterogeneous responses to political assassinations de-
pending on politiciansreactions to these events (see Davenport et al., 2019).
We document the moderating role of negative campaigning. Third, the case
study and the new data we collect allows for both causal identif‌ication and a
careful exploration of the proposed mechanism. Our approach, leveraging
short-run variation and high-frequency data, thus overcomes limitations of
previous studies that either rely on correlational evidence (e.g., Iqbal & Zorn,
2008), or estimate the effect of assassinations by comparing successful as-
sassinations to unsuccessful attempts (see Jones & Olken, 2009).
2
Fourth, we
contribute to the literature on the effects of negative campaigning by pointing
2088 Comparative Political Studies 55(12)

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT