Contextual effects, awareness, and voting behaviour: Does knowing about local politics increase contextual influence?
| Published date | 01 August 2019 |
| Date | 01 August 2019 |
| DOI | 10.1177/0263395717739860 |
| Author | Moreno Mancosu |
| Subject Matter | Articles |
research-article2017
Article
Politics
2019, Vol. 39(3) 315 –331
Contextual effects, awareness,
© The Author(s) 2017
Article reuse guidelines:
sagepub.com/journals-permissions
and voting behaviour: Does
https://doi.org/10.1177/0263395717739860
DOI: 10.1177/0263395717739860
journals.sagepub.com/home/pol
knowing about local politics
increase contextual influence?
Moreno Mancosu
Collegio Carlo Alberto, Italy
Abstract
Context is usually employed to explain voting behaviour, but the way in which it affects people
remains obscure. Previous studies have stressed that awareness of the context might indirectly
account for political environments’ influence. People aware of their context are expected to be
more affected by the relative prevalence of party supporters in their place of residence compared
with unaware citizens. By employing stacked logistic regression models (and using Italian National
Election Studies data), it is shown that people aware of parties’ settlement in their municipalities
are more likely to vote consistently with their surrounding political context.
Keywords
geography, Italy, political behaviour, political context, stacking
Received: 8th March 2017; Revised version received: 31st July 2017; Accepted: 28th September 2017
Introduction
Geographical context has been argued to play a prominent role in shaping citizens’ politi-
cal attitudes and behaviour. Several studies (e.g. Highton, 2000; Huckfeldt, 1979; Huckfeldt
and Sprague, 1987; Johnston and Pattie, 1998; Pattie et al., 2015; Pattie and Johnston,
2003; Putnam, 1966) have stressed that local geographical factors can affect individuals,
leading them to change their political ideas and beliefs. However, although largely
employed in the literature, the theoretical link between the place in which one lives and his
or her political attitudes has been poorly developed. Baybeck and McClurg (2005: 493)
define the context’s theoretical construct, as it is usually treated in the literature, as an
‘amorphous concept’. Consequently, theoretical arguments about the ways in which local
context affects individuals’ political behaviours have been poorly developed in the litera-
ture. The theoretical assumption – usually not completely explicit – behind the studies on
contextual effects is that geography can directly exert its effect on the individual by means
Corresponding author:
Moreno Mancosu, Collegio Carlo Alberto, Piazza Vincenzo Arbarello, 8, 10122 Torino, Italy.
Email: moreno.mancosu@carloalberto.org Twitter: morenomancosu
316
Politics 39(3)
of exposure to interpersonal influence among citizens belonging to a certain geographical
area (Baybeck and McClurg, 2005; Huckfeldt and Sprague, 1987), and indirectly by means
of political cues such as yard signs, lapel pins, and bumper stickers (Huckfeldt and Sprague,
1995).
Geographical effects, however, could be illusory and result from mere aggregation
biases (Baybeck and McClurg, 2005; Marsh, 2002) or homophily mechanisms (Bello and
Rolfe, 2014) rather than an actual influence of the geographical context.
The literature collected limited evidence for assessing whether a geographical location
actually affects its inhabitants (Books and Prysby, 1991). The most noticeable example of
a systematic attempt to test the mechanisms of geographical effect is the study by Baybeck
and McClurg (2005). By taking into account the awareness of the context that individuals
have, the authors have demonstrated that people who are more aware of the characteris-
tics of their geographical context of residence are also more prone to vote consistently
with their context compared to unaware people – namely, voting for the strongest party in
the local context. According to the authors, this could be intended as an indirect proof of
the fact that context matters: if context actually affects people’s choices, unaware citizens
will be less likely to be exposed to and influenced by it.
This article addresses this issue by employing data from Italy (in particular, using the
Italian National Election study (ITANES) of 2013, Vezzoni, 2014), while usually these
hypotheses are tested in the American context (with few exceptions, e.g. Pattie and
Johnston, 2000). Italy has a completely different institutional setting compared with the
United States, sharing nearly nothing in terms of historical backgrounds, party systems,
political participation, or mechanisms of democratic accountability. However, if the
mechanisms that link geographical context and political behaviour are general, the results
should be consistent in different political and geographical contexts.
Moreover, the Italian case is of particular interest since it presents a specific local
geographical structure. Since the end of the Second World War, the presence of two pri-
mary ideologies, Catholicism and Communism, placed, respectively, in the north-east and
the centre of the country has led to a long-lasting stability of voting preferences in those
regions (Agnew, 1996; Diamanti, 2003; Galli, 1968; Mancosu, 2016; Passarelli and
Tuorto, 2012). In 2013, the heirs of these two subcultures (the right-wing Popolo della
Libertà and the left-wing Partito Democratico) could count on strong support in their
respective zones of influence. Furthermore, the 2013 elections witnessed the appearance
of new political alternatives characterized by a harshly critical attitude towards the estab-
lished party system, the Movimento 5 Stelle – which in 2013, at its first appearance in a
general election, gained 25% of the valid votes and became the largest party in the coun-
try (Chiaramonte and Maggini, 2013; Vezzoni and Mancosu, 2016) – and Scelta Civica,
led by the former leader of the Government, Mario Monti. It is important to underline that
these two new parties produced, although in a more scattered fashion, something similar
to traditional parties’ ‘subnational spheres of influence’, both presenting clusters of
homogeneous support (De Sio et al., 2013). Geographical context and its effects, thus, are
particularly important – and are expected to exert a particularly strong effect – in the
Italian case.
Italy represents a specific context also from the methodological approach that we will
adopt: a multiparty system, like the one emerged from the 2013 national elections in Italy,
represents an issue wherein we are interested in testing a mechanism that does not involve
a single party. Indeed, we are not interested in how a favourable/unfavourable context for
a certain party modifies individual propensities to vote for it, but we are interested in a
Mancosu
317
mechanism that includes all relevant parties in the electoral competition. In this article,
instead of relying on multinomial logistic models that are relatively difficult to interpret,
we opt for the technique of ‘stacking’ (Van der Brug, 2004; Van der Brug and Mughan,
2007; Van der Eijk et al., 2006). Through the transformation of the dataset and individual
characteristic variables, stacking allows assessing the effect of a variable on a ‘generic’
party, leading to much more synthetic and readable coefficients.
The results show that aware and unaware citizens are split into almost two halves.
Moreover, it is shown that political sophistication of citizens and the ‘accessibility’ of the
context lead to more awareness of it: citizens are more easily aware of a place in which a
party is strikingly victorious compared with a place in which the first and the second par-
ties are engaged in a close race. The results corroborate our main hypothesis, showing
that the contextual effect is stronger among those who are aware of it. Finally, we assess
whether this result also holds at different levels of context accessibility, namely, whether
the differences in the local competition affect the differential between the aware and the
unaware. We show that aware people are not affected by the different levels of competi-
tion, while the contextual effect of unaware people decreases as long as the competition
decreases.
Background
The concept of ‘contextual effect’ is rather broad and obscure if we rely on the vast litera-
ture that has faced the problem of contextual influence (Marsh, 2002). Behind this generic
concept, however, it is possible to derive several theoretical arguments.
First, according to the literature, we can argue that the individual is exposed to the
geographical context in a probabilistic way (Baybeck and McClurg, 2005; Berelson et al.,
1954; Huckfeldt and Sprague, 1987). Given a certain area, as long as the prevalence of
party A becomes larger, it is increasingly likely that an individual will be exposed to party
A’s supporters and messages. Nevertheless, citizens do not interact with discussants they
pick randomly from the territory. People share a certain number of ‘social spaces’ with
their relevant others (Baybeck and McClurg, 2005; Huckfeldt and Sprague, 1995;
Mancosu and Vezzoni, 2017a), namely, formal or informal groups in which they can
socially interact – churches, workplaces, neighbourhoods, public places, and so on. The
theory of geographical context influence argues (usually implicitly) that social spaces are
partly affected by the prevalence of political preferences in the context surrounding them
(Baybeck and McClurg, 2005; Huckfeldt and Sprague, 1987). As long as party A is
stronger in a local context, the social spaces also included in this context will see...
To continue reading
Request your trial