As COVID-19 sends political campaigning to Facebook, will polarisation increase?

[Note: This article was co-published with OC Media, here. The article was written by Dustin Gilbreath, Deputy Research Director at CRRC Georgia. The views presented in the article do not necessarily reflect the views of CRRC-Georgia or any related entity.]

With Georgia in an election year and traditional face-to-face campaigning out of the question given the COVID-19 outbreak, the importance of Facebook in Georgian politics is only likely to grow.

Facebook is an important part of Georgian politics. Political campaigns are fought, and public opinion thought to often be formed on the platform.

The Government of Georgia and the ruling Georgian Dream party found it so important that they even set up numerous fake accounts posing as news sources, some of which Facebook later took down.

The perceived importance of Facebook is likely well-deserved. Among the 70% of Georgians that use the internet at least sometimes, it is by far most people’s most frequent activity; 72% of the public reports that one of their main three online activities is using Facebook, according to the NDI and CRRC November and December 2019 survey.

Given this, the question emerges, do Facebook users in Georgia have different political attitudes than non-users?

An analysis of the November and December 2019 NDI survey suggests that they have relatively similar attitudes to other internet users with one key exception — Facebook users have stronger opinions on political issues.

This suggests that if political campaigning moves further onto Facebook, people’s views could become more entrenched leading to more political polarisation.

On the November/December NDI and CRRC survey, respondents who reported using the internet were asked how often they encounter political news on Facebook. Among the response options was, ‘I do not use Facebook’, which 8% of internet users reported.

Based on this figure and a question on internet usage, a third (31%) of the public report not using the internet, 64% report using the internet and Facebook, and 6% use the internet, but not Facebook.

To understand who was more or less likely to be part of these different segments of society, a statistical model controlling for age, settlement type, household wealth, education level, and sex was run.  The data suggest that Facebook users and internet users who do not use Facebook are more demographically similar to each other than those that do not use the internet.

The results suggest that people who use the internet but not Facebook are less likely to live in urban areas outside Tbilisi, are older (average age of 47 versus 39), and more likely to be male.

Those who use the internet but do not use Facebook compared to people who do not use the internet live in wealthier households, are more likely to have a higher education, are younger (average age of 47 versus 60), and are more likely to live in urban areas outside the capital and rural areas than in Tbilisi.

When comparing those who use Facebook to those who do not use the internet, the pattern is similar.

Given the large role that Facebook plays in politics in Georgia, it would be reasonable to assume that people who use Facebook and people who do not but are still online might have different political views.

To explore this issue, a matching analysis was used to identify individuals that are similar along demographic lines, except for the fact that they either use Facebook or they use the internet, but not Facebook.

The results show few differences. The two groups do not have significantly different preferences for political parties. They both also tend to assess government performance similarly. They are equally likely to report that they are going to vote in the next parliamentary elections. They are also no more or less certain in who they are going to vote for.

There is one important difference, however — people who use Facebook are more likely to express their opinions. People who use the internet but not Facebook reported they don’t know and refused to answer questions significantly more often than people who use Facebook in this survey.


This finding has a number of potential interpretations. It may suggest that Facebook is informing people about politics in the country, and therefore, they can respond to the survey questions, which focus on politics, more easily.

It could also suggest that Facebook is polarising in Georgia. People that use the platform are significantly less likely to report uncertainty on the wide variety of issues asked about on the survey, hinting at stronger opinions.

Aside from these potential explanations, caution is warranted in interpreting Facebook as causing these patterns. Another potential interpretation is that people who do not use Facebook but are online are more cautious in sharing their opinions in public.

This would explain why they refused to answer more often and are not engaged in a platform that thrives on people sharing news about themselves and their views on politics. However, working against this view is the fact that both groups reported equal comfort in expressing their opinion in a quasi-public forum.

Taken together, the data suggests that there are relatively few differences between people who are on Facebook and not on Facebook but still using the internet, with one key distinction. People using Facebook are more likely to express their opinions.

This may point to Facebook either serving as a tool to inform the public or as a source of division. Alternatively, Facebook may draw the already more opinionated and informed. Potentially, it is both.

In either case, if politics is increasingly concentrated on Facebook in light of the COVID-19 outbreak, Georgian voters may become more informed and opinionated about politics. With stronger opinions, polarisation too may become stronger in Georgia.

Note: The data used in the above is available here. Replication code for the analysis is available here. In some cases in the above, figures may not sum to 100%. This is generally due to rounding error.