The Impact of Social Media Access on Political Support in Uganda
Social media has the potential to reduce support for elected autocrats by informing and engaging citizens. In response to some autocratic leaders limiting the public’s access to social media, researchers evaluated the impact of subsidizing access to social media on support for the ruling autocratic party in Uganda. Subsidized social media access reduced support for the ruling party among prior supporters by exposing them to perspectives critical of the government. Among non-supporters, subsidized social media access did not have any effect on the level of support.
الموضوع الأساسي
Social media makes it easier for people to share and access information, which can influence public opinion. At the same time, misinformation coupled with echo chambers reinforced by algorithms can entrench citizen’s prior beliefs. Evidence from established democracies suggests that social media primarily shows users content that aligns with what they already believe, so their political views do not change much. For example, a study conducted in the United States found that Facebook’s algorithms limit users’ exposure to posts that oppose their beliefs.1 But in elected autocracies—where leaders have more control over traditional media and people are less divided politically—social media might be able to change political opinions. Can subsidizing social media access decrease support for elected autocrats?
سياق التقييم
Uganda, where this evaluation took place, is an electoral autocracy that has been ruled by the National Resistance Movement (NRM) for close to four decades. Despite holding regular elections, Uganda is classified “not free” by Freedom House, describing NRM as retaining power through favors, threats, and unfair legal actions against opposition leaders.2 The government also exercises considerable control over traditional news media, which includes radio, television and newspapers. However, with increased internet usage—by late 2020, 52 percent of Ugandans had mobile internet access—social media is becoming an increasingly popular source for news. According to an Afrobarometer survey, 14 percent of Ugandans used social media to obtain news at least a few times a week, and 88 percent of users believe that it informs people about current events. In 2018, NRM reacted to opposition parties’ widespread use of social media by introducing the “over the top” tax (OTT) on daily social media use and imposed an internet blackout for over a month during the 2021 election.
معلومات تفصيلية عن التدخل
Researchers evaluated the impact of providing three months of subsidized social media access
on support for the NRM in Uganda. Researchers randomly assigned 1,389 individuals from 11
districts of Uganda to two groups: the treatment group received payments for the OTT tax and
mobile data bundles of 400-500 MB per week for the months of June, July, and August in 2021
whereas the comparison group received a smaller but flexible mobile money transfer of UGX
6,000 (US$1.65).
Surveys were conducted before and after the intervention to measure support for the NRM
and opposition parties in three different ways. Participants’ perceptions of party concern for
Ugandans, warmth toward parties, and openness to vote were measured and combined into an
index representing support for the NRM and opposition parties. These measures were then
aggregated to construct an index to represent support for the NRM and opposition parties.
النتائج والدروس المستفادة بشأن السياسات
Subsidizing social media access for three months negatively impacted support for the NRM
while positively impacting support for the opposition, but only for individuals who initially
supported the NRM. In contrast, non-NRM supporters did not update their support for either the
NRM or the opposition after the intervention, possibly because they had internalized the
criticism of the government before the intervention.
Social media usage: Individuals in the comparison group used around half of their transfer
toward social media on average. Researchers established that individuals in the treatment
group were 5.6 percentage points (a 35 percent increase) more likely to access social media
relative to those in the comparison group. Social media usage was measured by accessing
participants’ publicly available data on WhatsApp. WhatsApp usage correlated strongly with
self-reported use of Facebook and other platforms in the sample.
Political attitudes: Individuals in the treatment group who initially supported the NRM were 0.28
standard deviations less likely to support the NRM after the intervention. In contrast, these
individuals became 0.19 standard deviations more likely to support the opposition after the
intervention.
Individuals in the treatment group who initially supported the NRM were 8 percentage points
less likely (an 11 percent reduction) to believe that NRM cares most about them after the
intervention. Instead, these individuals became 9 percentage points more likely (a 37.5 percent
increase) to believe that the opposition cares most about them.
Individuals in the treatment group who initially supported the NRM expressed being 0.74 units
(measured on the scale of 0-10) less warm toward the NRM after the intervention. The
intervention had no measurable impact on these individuals’ warmth toward the opposition.
Individuals in the treatment group who initially supported the NRM were 0.26 units (measured
on the scale of 0-5) less likely to vote for a future NRM candidate after the intervention. The
intervention had no measurable impact on these individuals’ openness to vote for a future
opposition candidate.
Researchers found evidence that greater social media access reduced NRM support among
prior supporters despite not increasing their knowledge of current events. Reduction in NRM
support was also concentrated among those who had the greatest prior levels of faith in the
party. This suggests that the intervention reduced support among initial NRM supporters by
exposing them to perspectives on social media critical of the government.
Jeremy Bowles, John Marshall, Pia Raffler. “Access to Social Media and Support for Elected Autocrats: Field Experimental and Observational Evidence from Uganda.” Working paper, March 2025.
https://pubs.aeaweb.org/doi/pdfplus/10.1257/aer.20191777
https://freedomhouse.org/country/uganda/freedom-world/2025