The illustration shows the volume of tweets posted by users that were downloaded for each region in order to conduct the analysis.

The volume of tweets is certainly related to the population size of the regions themselves. It should be noted, however, that more tweets were downloaded and analyzed in Veneto than in Campania, albeit with a small gap, despite the fact that Campania possesses a larger population and is also the youngest region in Italy (it has the lowest old age and average age index of all the other regions).

​

A comparison of the time series of tweet volumes between the four different regions shows that there is a common peak of activity between Veneto and Campania on September 11 and an increase in tweets common to all regions on the 16th also of the same month.

In all regions analyzed around September 19, attention on twitter to the topic of regional elections begins to fade.

Hashtags most used by users in Campania:

Hashtags most used by users in Marche:

Hashtags most used by users in Puglia:

Hashtags most used by users in Veneto:

One of the most used elements on Twitter is definitely Hashtags. From the analysis of the latter, again from the users’ side, it can be seen that in Campania and Veneto the #s most used and with a large gap compared to all the others are those that refer to the candidates who later won the elections (De Luca and Zaia), while in the other two regions more generic #s related to the region and not to a particular candidate were used more frequently.

One of the most used elements on Twitter is definitely Hashtags. From the analysis of the latter, again from the users’ side, it can be seen that in Campania and Veneto the #s most used and with a large gap compared to all the others are those that refer to the candidates who later won the elections (De Luca and Zaia), while in the other two regions more generic #s related to the region and not to a particular candidate were used more frequently.

Cumulative tweet frequency of Candidates in Campania:

Cumulative tweet frequency of Candidates in Veneto:

Cumulative tweet frequency of Candidates in Puglia:

As done previously on the user side, we now turn to a temporal analysis of the volumes of tweets posted by individual candidates for each region.

In Campania, although he went on to win the election, De Luca has never been very active on the social network, probably due to an age-related fact, as opposed to his main rivals.

This is a focal aspect of the analysis as it highlights how despite the fact that De Luca is not personally active on Twitter he is nevertheless constantly at the center of many discussions. However, we do not notice any common trends in activity on the platform by the candidates themselves, the only aspect to note is a great deal of activity by Granato in the first few days examined.

As far as Puglia is concerned on the other hand, we note how the differences in activity volumes are not so dissimilar among the three main candidates, only Laricchia turns out to be more active than the other two, and how there are peaks in the use of the social network by all the candidates on the same days: between September 6 and 8 and between September 14 and 15. Fitto and Laricchia then show two other common peaks: on August 31 and September 4. It is also noticeable from the time series that Fitto completely stops his Twitter activity on September 19, a day when Emiliano also has a huge drop in volumes but does not completely stop his activity.

In Veneto, similar to what happened in Puglia, PD candidate Lorenzoni did not use Twitter much as opposed to his direct opponent Zaia, who instead made intense use of it.

The idea behind the graph was to use the volumes of tweets posted by users as a proxy for the number of votes by coalition, with the ultimate goal of assessing whether the opinion formed on Twitter reflected the election result. This analysis was carried out on hashtags, mentions and tweet text. In the graph we find on the x-axis the time variable, on the y-axis the cumulative sum of tweets by coalition.

As can be clearly seen from the graphs, the situation in Campania and Veneto, the two regions with the highest number of tweets downloaded and analyzed, seems from the outset extremely well defined with the candidates who then actually won at the polls firmly in the lead from day one. The different attitudes of the two candidates toward the social network should be emphasized: as highlighted earlier De Luca was not very active but still hit his electoral target, Zaia on the other hand won the election and maintained massive activity on Twitter.

Much less clear-cut situation in the Marche region where the winning candidate Acquaroli turns out to be in the lead from the beginning, but with a smaller gap over his direct opponents; the newly elected president of the Marche region is overtaken in terms of virtual votes by his M5S rival on September 21, the day on which voting actually ended.

In Puglia, the application of the algorithm did not lead to the same result at the ballot box as the analysis shows the winner, and by no small margin, center-right candidate Fitto who instead lost the election to his center-left rival Emiliano. Emiliano.