A new cross-regional dataset on media and electoral malpractice is released!

We are delighted to present one of the major contributions of our project – cross-regional data on electoral malpractice in Russia at the State Duma and Presidential elections 2007-2021. The data covers socioeconomic characteristics of Russian regions, as well as variables on TV and Internet coverage in each subject of the Russian Federation. The dataset contributes with the new variables on electoral coverage during both Presidential and State Duma electoral campaigns which capture the number of messages on elections in Russian regional media sources and the number of messages on electoral fraud. The dataset also encompasses electoral outcomes for political parties as well as basic socio-economic variables. Data covers four electoral cycles for the State Duma elections (2007, 2011, 2016, 2021) and three cycles for presidential elections (2008, 2012, 2018). More detailed variables on media coverage are to be added in the coming months. In this post, we visualize basic descriptive statistics on reporting electoral irregularities, their distribution across federal elections and regions as well as correlations with the main electoral variables: turnout, voting, and economic development. This post is written by Margarita Zavadskaya, Valeria Caras, and Elena Gorbacheva

Reporting fraud across federal elections: highest numbers in 2012

Most of the studies on how fraud is connected with political behaviour focus on election day or ex post violations, while we also include ex ante violations, i.e. malpractice that occurs on the eve of federal elections. Fraud rarely occurs spontaneously under autocracies and usually, implementation of fraud requires certain preparations in terms of amending the legal framework, excluding and suppressing opposition candidates and parties well before the election day. Since the direct measure of violations is even more challenging compared to detecting fraud on election day, we rely on media data that contain any mention of electoral malpractice during the election campaign. Our data allow one to take a first glance at the joint distributions and draw preliminary conclusions.

We collected our data on media messages on elections from the Integrum, the largest Russian mass media database. We searched in the regional mass media messages that contain mention of elections in general and fraud during the time period of six months before and one month after the election¹. Media include newspapers and magazines, official publications, archives of the leading national and international information agencies. We include only Russian-language media that are registered in the Russian Federation. The latter may affect the degree of censorship and style of politically sensitive information. On the other hand, in cases when federal elections coincide with regional or local elections, the media may be more open or cautious in covering electoral campaigns. Finally, spreading information during elections is a subject to the national legislature that regulates political advertising. Bottom line, we expect that regional media may have a tendency to under-report elections and, in particular, fraud and other violations.

Abundant flows of information that bombard citizens with messages that upcoming elections would fall short of being ‘free and fair’ may operate as a self-fulfilling prophecy and demobilize opposition voters from polling stations. How negative mentions of elections are spread across regions and time? 

Figure 1. The number of messages on elections and number of messages containing any mention of electoral malpractice (upper pane), the share of negative messages relative to the overall number of messages on elections (lower pane)

 

Source: GitHub

The early results inform us that Russian media outlets have been posting on electoral malpractice with a dramatic degree of variation across regions and types of elections – legislative or executive. In absolute numbers, the presidential elections of 2012 had the widest coverage and biggest share of negative expectations. There is no surprise as legislative elections in December 2011 spurred nationwide movement ‘For Fair Elections’ and ‘angry citizens’ remained alert in March 2012 to pre-empt further violations. The last State Duma elections received the most scarce coverage in domestic media – 23,767 messages in 2021 against 125,829 in 2012. Presidential elections, in general, draw more media attention, however legislative elections seem to attract negative messages more often. The biggest share of the messages containing information on fraud, electoral irregularities and violations is also observed in 2012 – more than 11% of all messages containing information on forthcoming elections. More than 8% of all electoral messages contained information on malpractice in 2011 and 6% – in 2021. These are elections with the most negativity relative to the total number of messages on elections. The least covered elections took place in 2016 with 4% of messages containing information of fraud. 

However, these data must be treated with a lot of caution. First, an important caveat is that news messages containing the keywords ‘fraud’ and their synonyms may not necessarily reflect negativity, they could also target the opposition or could even be devoted to the prevention of fraud. Second, the overall amount of information has a tendency to grow exponentially over time that making cross-time comparisons complicated. For instance, the latter implies that the overall scale of information has been growing over time as well as the very number of outlets keeps increasing. All this led to a bigger number of duplicated and shared news messages. This implies that the topic of elections and electoral integrity has been dramatically downplayed in the Russian media space.

Spreading information on electoral malpractice across regions: sources of variation

Media coverage of elections has been extremely uneven across regions and has changed a lot over time. Interactive maps show the proportion of news messages reporting electoral malpractice six months before the election day and one month after the national vote²: the darker the region’s shade, the higher the share of such messages from all messages on elections during the electoral campaign.

In 2021, more negative messages are observed in Ivanovo, Bryansk, Komi Republic, and Khabarovsk regions. The last two regions have been marked by protest activities: against Shiyes landfill construction³ in the neighbouring Arkhangelsk region from 2018 onwards in Komi and citizens’ mobilization supporting former governor of Khabarovsk Krai Sergei Furgal who was sentenced for alleged involvement in several murders. Among regions with the highest share of negative messages during the election campaign are Tuva Republic, Jewish Autonomous Oblast, and Astrakhan Oblast. Perm Krai, Tambov oblast, and the Chechen Republic demonstrated the lowest share of negative news messages during the last legislative elections. While Perm is known for relatively ‘clean’ elections and competitive regional politics back in the 1990s and early 2000s, Chechen Republic is the most closed and authoritarian region with tight censorship and marred with violence. Thus, the mechanisms that drive the observed values differ dramatically. Leaders in terms of an absolute number of messages in 2018 and 2021 – Moscow, Saint Petersburg, Sverdlovsk oblast, and Yaroslavl’ oblast. These regions are known for vivid civil organizations and the presence of regional election monitors. Below is the graph representing the number of messages about elections and electoral fraud for Moscow. Moscow proved out to be an outlier case due to significantly larger numbers of messages, thus we decided to put it on a separate graph.

Figure 2. The number of messages on elections and number of messages containing any mention of electoral malpractice for Moscow.

Source: GitHub

Higher frequencies and shares of negative messages concerning upcoming elections may result from a variety of sources: 1) the presence of election monitors and civil society groups that are well organized and enjoy sufficient capacity to collect information of existing or possible violations, 2) a more competitive regional political landscape and presence of more active opposition and independent groups and media, 3) vice versa, genuinely repressive regions that systematically manifest the worse quality of elections, 4) concurring regional and/or local elections that may increase turnout. Finally, all these mechanisms may operate simultaneously and it remains a research agenda for further studies.

Figure 3. Share of messages on electoral fraud from the total amount of messages on State Duma elections 2021 ranged by regions.

Source: GitHub
Some of the regions are absent from the graph due to the fact that there was no fraud mentioned in mass media. Various factors can explain the absence of coverage of electoral fraud and they may vary from region to region.  For instance, one reason why Kurgan oblast’ is absent from this graph may be explained through the independent local media crackdown in 2020, as Kurgan regional representative of Golos Mikhail Kutuzov mentions.

The impact of e-voting

In 2021, electronic voting was implemented for the first time in the federal elections. Among the regions where it was used are Moscow, Murmansk, Kursk, Nizhny Novgorod, Rostov, Sevastopol, and Yaroslavl. Three of them – Moscow, Nizhny Novgorod and Yaroslav – have been at the top of the rating in terms of negative messages on elections before 2021 and still remain there (see Graph 6). Rostovskaya oblast also scores quite high on the scale of negative messages, while Kursk, Sevastopol and Murmansk remain in the lower part of the bar chart. At the first glance, there is little connection between e-voting and the scale of negative messages on fraud.

Effects of Domestic Monitors

Russian domestic election monitor Golos are represented in about half of Russian regions (44 out of 85) which implies that they improve the coverage of electoral malpractice and make it visible to a wider audience including the media. However, domestic monitors were present in none of the regions from the top-5 with the biggest numbers, but Ivanovo oblast. Previous elections were marked with the elimination of independent observers and media representatives from the polling stations. For instance, some experts call Bryansk (the region with the second-highest share of electoral fraud messages in 2021)  ‘an electoral sultanate’ recalling an infamous incident when at the polling stations with even numbers the election commission put the turnout of 91% and with those with odd numbers of 90%. 

Media and Political Participation: A Weak Demobilizing Effect?

The scatterplot in Figure 4 shows a non-linear negative association between official turnout and the share of negative media messages in each region in 2021. We observe a weak logarithmic correlation between turnout and logged share of negative messages (Adjusted R2 = 0.05, p = 0.02). Each additional message containing information on fraud is associated with lower turnout.

Figure 4. Relation between the turnout and the share of negative media messages about the 2021 State Duma elections in each region of the Russian Federation.

Source: GitHub

Moreover, bigger numbers of messages containing information on actual or potential electoral fraud are systematically associated with a higher share of votes for the CPRF (Figure 5). Although, we find no correlation between the vote share for the UR and media messages (see Figure 6). National republics (with the exception of Komi) tend to have higher support for the UR regardless of the reported fraud in the media. Ivanovo oblast’, which demonstrated the lower share for UR and higher share of negative messages has a developed network of electoral observers. We assume their presence made the fraud more exposed to a broader audience and amplified its media coverage. It must be noted that Ivanovo and Bryansk regions were mentioned by the head of the Central Election Commission Ella Pamfilova as one of the regions with electoral violations.

Figure 5. Relation between the CPRF votes and the share of messages on electoral violations during the 2021 State Duma elections in each region of the Russian Federation.

Figure 6. Relation between the United Russia votes and the share of messages on electoral violations during the 2021 State Duma elections in each region of the Russian Federation.

Source: GitHub

A cursory glance at the interconnections between media coverage and electoral outcomes demonstrates that even in authoritarian states there is a wide variance. However, the nature of causal relations remains unclear: is it that more competitive and pro-opposition regions tend to have a more even coverage of elections or frequent mentions of fraud re-shape the agenda and draw more attention from voters? We will explore these and other questions in the upcoming articles.

---

¹ In the next version of the dataset which we will publish later this year, we will divide the messages into three groups – published before the voting day(s), during the voting day(s), and afterwards.

² In 2021, State Duma elections were held within three days.

³ Komi Republic shares the border with Arkhangelsk oblast and Shiyes environmental protests spread to Komi as well.