How SMS texts on corruption helped to empower Ugandan voters at the polls

Uganda, in eastern Africa, is notoriously corrupt, and many Ugandans have little faith that their leaders will provide them with honest information. But a new study conducted in the country suggests that mobile telephone technology and official information about budget corruption can help citizens hold politicians accountable at election time.

Published in the Proceedings of the National Academy of Sciences, the study details how Uganda's ubiquitous mobile phones and targeted SMS text messages about budget irregularities were able to help voters make better informed choices at the polls.

Mark Buntaine, an assistant professor in UCSB's Bren School of Environmental Science & Management and one of four co-authors of "SMS texts on corruption help Ugandan voters hold elected councillors accountable at the polls," noted the 2016 experiment focused on district elections. Politicians at this level are directly responsible for basic public services such as trash hauling, water wells and the like.

"This is really the front line where people experience the things that they care about," he said, "and if they can get accountability at that level, there's a lot to be achieved."

The study involved roughly 16,000 voters, all of whom agreed to participate. Working with Twaweza, a Ugandan-based organization that promotes good governance, the researchers used mobile phone text messages to inform voters about suspected budget fraud by local government councils.

Allegations of suspected fraud were based on annual government audits that track whether councils followed procurement rules, completed projects as specified and properly accounted for expenditures.

The results showed that voters who confirmed receiving messages and who learned that suspected fraud was greater than they expected were 6 percentage points less likely to vote for incumbent councilors. Meanwhile, voters who learned that fraud was less than expected were 5 percentage points more likely to vote for incumbent councillors.

The same messages had no effect for higher-profile elections for council chairperson, the study found.

"Sending factual information works some of the time," Buntaine explained, "for lower-level politicians who are more like city councilors, the first level of people you would contact as a citizen. Once you get up higher on the chain, we don't observe any effect on voters' choices. We think that's because voters have more information about higher-level politicians, so there's less room to move people's choices at that level."

Buntaine cautions that the study should not be seen as a template for interventions in other elections. The Uganda experiment used information that was "salient, local and attributable to specific kinds of politicians. If you lose any of those characteristics, you likely lose the ability to change people's minds about how to vote," he said.

Indeed, the Uganda study was part of a larger initiative, Metaketa I, in which research teams study the effect of information on voter behavior in six other countries. The results in Uganda were unique. "Our finding contrasts with other studies and even other types of messages in our own study that produce a null result," Buntaine said. "Information doesn't seem, on average, to change the way people vote."

Nonetheless, he said, just because an approach on average doesn't work does not mean that it never works. "We need to focus on the conditions that allow for information to have a beneficial effect on the way that people approach elections. And again, we think that's when it's salient, it's local and it's very attributable to politicians."