December 9, 2020

Reflections on Data + Elections (by Rob Macdonald)

Many commentators highlighted the potential for voters to be empowered by the growth of mobile telecommunications across the African continent. However, this initial enthusiasm has somewhat waned as accounts of targeted messaging during elections have become common, particularly as the content of these messages has often been misleading, out-right false, or inciting. The recent rise of online campaigning through relatively new social media platforms has raised further concerns about how the required data is obtained, the extent to which African democracies are vulnerable to foreign interference, the ways in which social media algorithms are prone to manipulation, and the ethics of using African countries as a testing ground for new digital technologies.

These were some of the themes discussed during the latest Developing Data workshop on Data + Elections on 25 November 2020. The session was chaired by the project’s Principal Investigator Thomas Molony.

The first panellist was Laura Lazaro Cabrera from Privacy International. She gave a broad overview of the data exploitation in electoral campaigning ecosystem. It includes three main sources of data: electoral registers, canvasing, and third parties/data brokers. The third parties’ data can include things such as transactional data, consumer surveys, and behavioural data about online activity. Lazara Cabrera then explained that data is increasingly used by political parties for micro-targeting in a four-stage process. Firstly, the data is collected, then individual voters are profiled into small groups based on their attributes. Next, advertising material is personalised and, finally, personalised content is distributed on social media platforms to reach the targeted group.

Lazaro Cabrera also outlined some safeguards against data exploitation, such as data protection laws. Where they exist, they can require consent for data processing or give voters the right to ask political parties about the information they hold. In some cases, regulators will set further limits to the ways in which data can be used for micro-targeting. However, due to loopholes, data can often be used without consent.

The next panellist was Seema Shah of The Africa Centre for Open Governance. With a focus on both traditional and social media, she examined how big data can help actors in the political space understand and predict voter behaviour and facilitate targeted messaging. She then discussed how the associated reliance on algorithms is problematic, not just because they can be wrong, but also because, by making people into numbers, they downplay the understanding of people as people rather than at an aggregate level. The importance of context can be lost and citizens whose data is not represented in the data, because they do not have smart phones or social media, can be left out of the calculations political parties are making.

Next Shah gave the example of Cambridge Analytica taking over the branding of Jubilee’s campaign during the Kenya 2017 elections, and the misinformation that was circulating at that time. This led to her raising several broader concerns about micro-targeting. She pointed to concerns over privacy, the formation of online bubbles leading to people taking increasingly extreme positions, the undermining of media legitimacy, and the spread of misinformation undermining voters’ ability to make good decisions in elections. Ultimately, she argued, the companies employed to run micro-targeting initiatives are interested in profits not debating ideas or helping citizens and this is bad for democracy.

The final panellist was Boye Adegoke of Paradigm Initiative. He discussed digital rights in Nigeria. Although social media, such as Facebook had been used in conventional ways earlier, Adegoke argued that the 2015 elections in Nigeria were the first in which people became aware of the power of data. Cambridge Analytica was also present at that election, and it attempted to use religious tension to cause disaffection among certain sections of the electorate. However, as Adegoke explained, in this case the narrative was successfully countered.

Adegoke gave some examples that show political data is being marketed in Nigeria. In 2016, information about registered voters was put on sale by a broker despite their being no mechanism through which the electoral commission should have been able to release it. There have also been political cold-calls to mobile phones in which the callers refuse to disclose where they got the numbers. Although Adegoke argued that there is yet to be an effective use of social media during a Nigerian election, he pointed to weak data governance when he suggested that the scene is set for data to make a significant impact at future elections.

In the question-and-answer session a few further insights were brought out, including:

  • There is still limited evidence as to how effective micro-targeting is in changing voter behaviour
  • The issue of micro-targeting is still not well understood by broader electorates.
  • One of the major reasons that states are reluctant to pass stronger data protection legislation is fear that it will be used against them.
  • Much of the data exploited during micro-targeting is sold by low-level employees of electoral commissions and companies without authorisation.
  • COVID-19 has implications for data and elections as it means voters will spend more time on digital devices and that contact tracing apps provide a new source of data.

Leave a Reply

Your email address will not be published. Required fields are marked *