Recurrent patterns of user behavior in different electoral campaigns: A Twitter analysis of the Spanish general elections of 2015 and 2016

TL;DR

It is revealed that the daily number of tweets, retweets, and mentions follow a power law with respect to the number of unique users that take part in the conversation.

Abstract

We have retrieved and analyzed several millions of Twitter messages corresponding to the Spanish General elections held on the 20th of December 2015 and repeated on the 26th of June 2016. The availability of data from two electoral campaigns that are very close in time allows us to compare collective behaviors of two analogous social systems with a similar context. By computing and analyzing the time series of daily activity, we have found a significant linear correlation between both elections. Additionally, we have revealed that the daily number of tweets, retweets and mentions follow a power law with respect to the number of unique users that take part in the conversation. Furthermore, we have verified that the topologies of the networks of mentions and retweets do not change from one election to the other, indicating that their underlying dynamics are robust in the face of a change in social context. Hence, in the light of our results, there are several recurrent collective behavioral patterns that exhibit similar and consistent properties in different electoral campaigns.

Figures (11)

• Figure 1: Scheme that shows the method used to build the networks of retweets. In panel 1), user $i$ posts a message starting an information flow that reaches user $j$, who decides to retweet it. In panel 2), the retweeted message posted by $j$ is read by $k$ who in turn retweets it again. In panel 3) the three users are information transmitters and in the last panel the resulting retweet network is shown.
• Figure 2: Left panels: time series of aggregated user activity per day for the 2015 (top) and 2016 (bottom) elections. The shadowed region corresponds to the days of electoral campaign. Right panel: Linear regression between the activity time series of both elections. The lines correspond to the linear fits of the data including the day of the elections (red dashed line) and excluding it (black continuous line).
• Figure 3: Power law relationships of the total number of tweets, retweets and mentions per day as a function of the number of unique users that participated in the conversation each day for the 2015 campaign (top) and the 2016 campaign (bottom). Note that the data corresponding to the day of the elections (marked with a circle) were not included in the fit.
• Figure 4: Temporal evolution of the distribution of activity per day in both electoral campaigns (2015 in the top panels and 2016 in the bottom panels). Left panels: probability mass functions (PMF) of the distribution of activity for each day in color code. Right panels: daily evolution of the $\gamma$ exponent of the power law fit of the activity distributions. The error bars correspond to $2 \sigma$.
• Figure 5: Left panel: strongly connected component of the aggregated mention network for the 2015 electoral campaign. Right panel: strongly connected component of the aggregated retweet network for the 2015 electoral campaign. Colors correspond to the communities computed with the Louvain algorithm 1742-5468-2008-10-P10008. We have indicated the most probable affiliations of the nodes of each community by visually inspecting which users correspond to the most central nodes. Every well defined group seems to correspond to a political party. The size of the nodes is proportional to $\log(PageRank)$.