On December 6th Venezuelans are once again heading to the polls, this time around to choose the members of the National Assembly for the 2016-2020 period. Polling suggests that for the first time since 2007, the government is set to lose. However, because of an electoral system with a significant rural-bias, the opposition’s large lead in polls (about 20%) might not translate to a majority of equal size (see an analysis here).
The electoral campaign started officially on Friday the 13th of November. Already before that date the government was working on adapting the system to its advantage. With two weeks to go before the election, the environment is heating up. As has become common in recent elections in Venezuela, electoral violence has started occurring. Two key incidents have happened within the last week, one in Petare and the other in Guárico.
The increased polarization of politics in the past seventeen years, together with the lack of a viable path to power for the opposition and an accommodation in power on the side of the government, have made electoral contests fields of battle instead of routine democratic events. For instance, former president Hugo Chavez was always quick to link every election with a military battle. In the current election, President Maduro and government supporters have coined the phrase #SeaComoSea – “whatever it takes” (to win the election). On the side of the opposition some leaders are talking about a new Venezuela on the day after the election #PaseLoQuePase – “whatever happens”. These phrases certainly don’t help in ameliorating the tense situation in Venezuela ahead of the election. It is in this context that there are more instances of electoral violence at each election. The killing of an opposition party leader in the state of Guárico this past Wednesday signals an increase in the violence caused by armed groups ahead of elections.
Growing research in the field of electoral violence has found that incumbents who fear to lose elections are more likely to use violence (Bashin & Ghandhi, 2013; Hafner-Burton et al 2013). While this can be seen as obvious, the research has found that this violence is typically: 1) aimed at opposition leaders rather than at the population as a whole; b) more likely to happen in the period right before the elections – rather than after or much earlier; c) more likely in close elections or when the incumbent is polling unfavourably; and d) carried out by a myriad of agents that range from military to para-military, to temporarily recruited youths (Bashin & Ghandhi, 2013; Hafner-Burton et al 2013).
As can be seen, the current situation in Venezuela mirrors many of the findings of the literature. The absence of many of the traditional electoral observers (EU, OAS, Carter Center), the rising tensions, and the probability that the government will not only lose but do so very badly all mean that the chances for more violence incidents are higher than in past elections. This is why at the University of Glasgow the research team on Explaining and Mitigating Electoral Violence has been following the situation in Venezuela by tracking Venezuelan social media.
On Sunday November the 22nd the monitoring team noticed a significant increase in Twitter activity, from 4000 tweets per hour to 12000 tweets per hour; a significant increase also happened on the 24th and 25th of November. The tweets monitored are from Twitter Streaming API using carefully selected keywords related to the Venezuela Election. In addition, we are also tracking relevant Twitter accounts of Venezuelan politicians from both the PSUV (government) and the MUD (opposition). In Figure 1, which shows the number of tweets collected each hour, the trend is clearly visible as is the abnormal increase.
Figure 1 Tweets Collected per Hour
While an abnormal increase in tweets alone is not necessarily indicative of anything in particular, tweets are also monitored for their content and in particular how they match pre-determined electoral violence keywords. Figure 2 shows the evolution of these tweets; notice that while there was an increase on Sunday the 22nd, the biggest increase in this figure occurs on the 25th of November. These two dates coincide with the situation in Petare and the killing in Guarico mentioned above. It is interesting to note the different attention the two events received; this is probably explained by the fact that the death of Luis Diaz is a much more severe incident of electoral violence. Not only was the incident a fatality, but one that took place on a public stage and was widely covered by the national and international media. The slight tail of tweets related to violence on the days following the 25th is probably due to the aftermath of the event.
Figure 2 The evolution of tweets matches pre-determined electoral violence keywords
Finally, examining the content of the tweets we can map the semantic patterns in them. Figure 3 shows a wordcloud that highlights the content of the tweets collected. The bigger the word the more times it gets mentioned. As it can be seen, the words Venezuela, National (Nacional), Assembly (Asamblea), and elections (elecciones) are some of the more frequently mentioned terms; however, these are closely followed by violence (violencia), murder (asesinato) and death (muerte). This shows that in the period from Saturday 21st November to Thusday 26th November, there were a significant number of tweets focusing on these electoral violence events.
Figure 3. Tweet content
In comparative perspective, the Venezuelan violence appears anomalous. Recent research by Van Ham and Lindberg (2015) has argued that incumbent governments choose between different manipulation strategies ranging from electoral violence to vote tampering. The choice of which of these strategies to use is dependent in part on the level of democracy of the country. In essence governments weigh the costs and benefits of manipulation (Birch, 2011). The interesting thing about electoral violence in the Venezuelan setting is that the country is neither poor, nor is it a nascent democracy at particularly low levels of democratic development. In this type of setting the theory would predict that fraud and vote buying would be prevalent. It seems though that because 1) the actual voting machines are difficult to tamper with; 2) there is a significant gap in the polling between the government and opposition; and 3) the regime is built upon the popular support it’s had in the past; then the costs of outright fraud are too high and as such electoral violence seems to be the way for confronting an unfavorable electoral situation.