Can Europe Make It?

The Italian electoral campaign and the issue of immigration: a Twitter analysis in the aftermath of the Macerata attack

Interestingly enough, the presence of hashtags such as #itaexit and #noEU suggests a clear link – in the ‘national closure’ discourse – between anti-immigration and anti-European sentiment. 

Giovanni Pagano Martina Zaghi Francesca Arcostanzo
26 February 2018
lead lead

Luigi Di Maio, Five Star Movement (M5S) leader, attends the presentation of the political program for the upcoming general elections in Rome on January 30, 2018. The Italian general election will be held March 4, 2018. Giuseppe Ciccia/Press Association. All rights reserved.In the run-up to the general elections on March 4, Italy has seen a rise of right-wing parties in the polls: the anti-immigrant Northern League, Silvio Berlusconi’s Forza Italia and the far-right Brothers of Italy are now heading into the poll with the winds at their backs. As these forces share a more or less pronounced anti-immigration stance, it shouldn’t be any surprise that immigration has been one of the most debated issues in the electoral campaign.

Recent events have contributed as well to make the issue of immigration even more prominent. For about three weeks now, the usually quiet town of Macerata has found itself at the heart of a heated – yet bitter – national debate about immigration. On January 31 a Nigerian man – Innocent Oseghale, whose residence permit had expired – was arrested on suspicion of the murder of Pamela Mastropietro, an 18-year-old Italian girl whose dismembered body was found in two suitcases in the environs of Macerata. On the morning of 3 February, Luca Traini – a far right extremist and onetime candidate for the Northern League – was arrested after wounding six African immigrants in a two-hour shooting rampage in the streets of the central Italian town. At the moment of the arrest, Traini wrapped himself in an Italian flag and gave a fascist salute. Italian authorities claim to have seized a Celtic cross flag (a symbol usually linked to neo-Nazism and white supremacism) and right-wing literature including Hitler’s Mein Kampf from Traini’s home.

Such a sequence of events turned up the heat of an already intense electoral debate. Northern League’s leader Matteo Salvini condemned the shooting, but accused the center-left of “flooding Italy with immigrants” and transforming the country “into a refugee camp”. Silvio Berlusconi – whose party Forza Italia is a major ally of Salvini’s League – referred to immigration as to a “social bomb”. Meanwhile, the center-left Democrats – the main incumbent party – are having a hard time finding a balance between firmly condemning the rising tide of xenophobia and reassuring citizens’ fears about immigration.

The debate on Twitter

To capture the reactions to the Macerata shootings and how the event has shaped citizens’ attitudes towards immigration, from January 27 to February 9 we have been following the Italian debate on immigration on Twitter. All in all, we collected more than 800,000 tweets - sent by almost 115,000 users – containing one or more keywords commonly used in the wide-ranging debate about immigration and/or the murder of Pamela Mastropietro and the Macerata shooting.

As can be seen in Figure 1, immigration issues are widely discussed on Twitter. As the electoral campaign heats up, the debate already runs at a constant but significant pace before the shooting in Macerata. On the other hand, such a traumatic event triggered a surge of online reactions, clearly visible in the ‘spike’ in Twitter activity centered on 3 March. Against the background of the political turmoil generated by the shooting, tweets about immigration reached a peak. The volume of conversations remained high for the following days.

Copy of Figure_1.png

‘Openness’ vs. ‘closure’ 

To better understand the ways in which Italian Twitter users have framed their discourse on immigration in the aftermath of the Macerata attack, we tried to bring to the surface tweets explicitly expressing attitudes or opinions, narrowing our focus only to conversations that took place after March 3. By means of natural language processing (NLP), we built an algorithm to separate ‘opinion’ tweets from news recast or descriptive tweets. In this way, we were left with more than half a million tweets (sent by over 80,000 users) expressing a political opinion or an emotional attitude towards immigration or immigration-related issues.

We then took a closer look at those conversations. Although the debate includes many different sub-issues, a clear polarization emerges between those users who tweeted blaming the attacks on the rising number of immigrants in the country, and those who took up to Twitter to condemn instigations to xenophobia. We processed our ‘opinion’ tweets through an algorithm to disentangle conversations and discussion that shape what can be illustrated as a divide between ‘openness’ and ‘(national) closure’, and to determine on which side of this divide Twitter users stand. Results of our classification are shown in Figure 2.   

Copy of Figure_2.png

Fig. 2. Share of pro-openness (55%) and pro-closure (45%) tweets in the aftermath of the Macerata Attack (February 3 - February 9 2018). As the chart shows, in the aftermath of the Macerata attack reactions from the ‘openness’ field seem to prevail over the ‘closure’ front, as the majority of opinions and attitudes in our tweets (55 percent) can be attributed to the pro-immigration area. To illustrate the many nuances of the debate, we used a topic modelling algorithm to identify distinctive features in each field and highlight different topics and discourses. The word-clouds in Figure 4 show the most characteristic features in each front.

Copy of Figure_3.png

Fig. 3. Most distinctive words and hashtags in ‘pro-openness’ (on the left) and ‘pro-closure’ tweets (on the right). The bigger the word, the higher the distinctiveness.As the clouds show, the fact that Traini’s drive-by shooting was not only racially motivated, but also fascist-inspired ended up influencing the debate, particularly on the ‘openness’ front. In the blue word-cloud, features such as fascism, anti-fascism, stop fascism, racism, or roman salute are the most evident. Ultimately, such an anti-fascist rhetoric appears as the most distinctive trait of the ‘open’ front, somewhat overshadowing arguments of solidarity and claims in favour of better integration policies. On the other hand, users in this field also tend to blame Salvini’s Northern League for the deterioration of the public debate, and for using the migrant crisis to foster xenophobia for political gain.

The red (closure) cloud is filled with terms that reflect an anti-foreigner sentiment: in the conversations we analyzed, words like drug dealer, immigrant, Nigerian, visa, deportations and so on, contribute to emphasize the tie between the migrant invasion and Italy’s various problems. Many users in this field blame the lax immigration policies of the centre-left government (Minister of the Interior Marco #Minniti) and the unintended consequences of the well-meaning rhetoric of humanitarian aid (championed by President of the Chamber of Deputies Laura #Boldrini). Interestingly enough, the presence of hashtags such as #itaexit and #noEU suggests a clear link – in the ‘national closure’ discourse – between anti-immigration and anti-European sentiment.

The political partisanship of anti-immigration users: a tentative estimation

Detecting the political partisanship of social media users is not a trivial task. Contrary to data collected from public opinion surveys – where respondents are often required to provide socio-demographic information, – social media data most often come without explicit information regarding users’ voting preferences. However, in recent years social scientists have been trying to estimate the political affiliation of social media users using a variety of methods based on users’ interactional behaviours on Twitter. In particular, scholars agree that ‘retweets’ can be considered as consistent indicators of support for a tweeter’s account and content (Conover et al. 2011a; Conover et al. 2011b).

Starting from this assumption, having identified the ‘anti-immigration’ users present in our dataset, we then attempted to estimate their political partisanship on the basis of their retweets during the last months. Of course, such a method does not allow for a precise determination of the political affiliation of all our users – which goes beyond the scope of this article – but we believe it can still provide useful insights into the relationship between citizens’ political preferences and their attitudes towards immigration.


More specifically, we wanted to check whether users who tend to retweet the leaders of parties displaying a negative attitude towards immigration (Northern League, Brothers of Italy, Forza Italia and, to some extent, the 5Stars Movement) were more likely to position themselves on the ‘closure’ front in the aftermath of Macerata terrorist attack. To do so, we collected all retweets made in the last three months by the 79,480 users who have tweeted their opinions about Macerata between February 3 and February 9. Among those, 4,972 resulted to have retweeted at least once contents from Matteo Salvini (Northern League), 2,605 from Giorgia Meloni (Brothers of Italy), 1,858 from Silvio Berlusconi (Forza Italia) and 2,478 from Luigi Di Maio (5SM).

As a second step, we looked at how these ‘partisan’ users have positioned themselves in the Macerata debate. As Figure 4 shows, supporters of right-wing parties tend to take anti-immigration stances even when debating the Macerata attack: 68% of the tweets sent by Giorgia Meloni’s supporters have been classified by our algorithm as pro-closure, followed by Matteo Salvini’s (67%) and Silvio Berlusconi’s (62%). 

Copy of Figure_4.png

Fig. 4. Percentage of ‘pro-closure’ tweets sent in the aftermath of the Macerata terrorist attack (3-8 February 2018) by users who in the past two months have retweeted at least once Giorgia Meloni (68%), Matteo Salvini (67%), Silvio Berlusconi (62%) and Luigi Di Maio (59%).The case of the 5SM is particularly interesting. As commentators point out, the 5Stars Movement is “a political construction unlike anything else in Europe. It has a rightist facade, over a leftist basement, under an anarchic roof”. Unlike other populist movements, the 5SM doesn’t fit neatly into the traditional far-left or far-right ideology, and its electorate is distributed along the left-right political spectrum, with a tendency towards the political centre. For these reasons, the 5Stars Movement has been very cautious in taking stances with regard to ‘divisive issues’ such as the common currency and immigration. While the 5Stars’ leadership has often been accused of building on widespread negative feelings about immigration, its electorate has proved much less cohesive on this issue than on other topics, such as socio-economic issues and the fight against political corruption.

Ultimately, the party stance on immigration is rather ambivalent and not always easily decipherable. Our results show that 59% of the tweets sent after Macerata by the 2,478 users who have retweeted Luigi Di Maio in the past two months have been classified as ‘pro-closure’. Although the percentage is significantly lower than that of ‘traditional’ right-wing parties, this finding suggests the existence of at least a group of Twitter users that, besides sympathising with the 5Stars Movement, share their view on immigration with right-wing parties and voters.

As the electoral campaign intensifies, political heat surrounding immigration remains high. Whatever the outcome of 4 March election will be, the issue of immigration is here to stay.

Get weekly updates on Europe A thoughtful weekly email of economic, political, social and cultural developments from the storm-tossed continent. Join the conversation: get our weekly email


We encourage anyone to comment, please consult the oD commenting guidelines if you have any questions.
Audio available Bookmark Check Language Close Comments Download Facebook Link Email Newsletter Newsletter Play Print Share Twitter Youtube Search Instagram WhatsApp yourData