by Amber Macintyre
In 2018, the Cambridge Analytica scandal brought worldwide attention to the use of voters’ personal data in political campaigns. The scandal helped spotlight the impact of digital campaigning firms, in particular in the US and the UK. Few of the headlines, however, addressed how such firms based in the US and Europe have also consulted for various political parties, using similar data-driven methods of influence, in other areas of the world including Sub-Saharan Africa. Yet the involvement, and consequences, can be substantial — in 2017, the founder of the Washington-based digital campaigning firm Aristotle International Inc., John Aristotle Phillips, was arrested and deported while his company was advising Kenyan opposition candidate Raila Odinga’s campaign. Meanwhile, Harris Media, a Texas-based firm, worked for Raila’s opponent, Uhuru Kenyatta, promoting a website showcasing President Uhuru Kenyatta’s accomplishments and producing an attack campaign on Raila Odinga through another website called “The Real Raila”.
US-based digital campaigning consultants [battle] each other in elections in African countries
The 2017 election in Kenya is just one instance of US-based digital campaigning consultants battling each other in elections in African countries, in which they either test out techniques in order to bring them back to the US or export digital services developed in elections in the US. Yet little in-depth research or investigation has been carried out into their impact on politics in Sub-Saharan Africa. Digging deep into the data-driven practices of political campaigns in these regions can help researchers, policy-makers and practitioners understand which practices are unique to the region, and which ones are used globally, as well as what consequences these practices have for political participation.
In Summer 2020, Tactical Tech’s Data and Politics team held a roundtable on the use of personal data in political campaigns in Sub-Saharan Africa, in collaboration with political analyst and author Nanjala Nyabola. The two-day, twenty-person discussion was held online with researchers, lawyers, communication professionals, policy-makers, campaigners and educators based in Uganda, Nigeria, Kenya, The Gambia, Zimbabwe, Togo, Ghana, Canada, UK and Germany. The discussions covered the data-driven tactics used by political parties, the actors they work with, the short- and long-term impact of data-driven campaigning on political participation, and what can be done to address any negative impact. The roundtable discussion brought to light unique insights about the specifics of how citizens’ personal data is leveraged for political ends throughout Sub-Saharan Africa.
Two of the most common data-driven methods used to reach voters and citizens in the region are bulk SMS messages and robocalls. Participants raised the issue that any methods reliant on phone numbers are common because they are easy to obtain — often through methods which could be considered illegitimate. For example, communication and internet payment companies are a major supplier of personal and political data, whether paid for or acquired through more questionable methods. One such company, Safaricom, are a government-owned communications utility in Kenya and have around 63% of the market share. The company gathers personal data for tax and other government purposes. Those using the network receive more politically-based text messages than anyone else. In 2018, after a text was sent to 10 million people in Zimbabwe on the morning before an election encouraging support for the ruling party, Zanu PF, and the president, Emmerson Mnangagwa, the largest communications and payment companies, EcoCash and EcoNet, were accused of being involved in exchanging phone numbers with the ruling party. Many participants commented that part of the reason this happens is due to the overuse of the justification of national security by governments to collect data from the communication companies.
Other legally and ethically ambiguous methods of data collection are also used to gain access to phone numbers. In Kenya, for example, paper sign-in sheets are used to register who enters and leaves buildings such as shopping centres or offices. Visitors have to leave their name, signature and often their phone number. The information can be collected by anyone who comes in to acquire them, likely for money, and then can be sold to other buyers, including political parties or their private campaign consultants and firms. One participant mentioned the private firm Eskimi, which allegedly have the most granular data in Nigeria and Uganda. In both countries, it is mandatory to register phone numbers alongside personal information, which makes it easier for private data brokers to connect names, addresses and phone numbers, to other personal information such as online browsing habits, financial transactions or location data.
It is easier in Sub-Saharan Africa to collect data through illegitimate, and at times, illegal, methods
These methods of data collection in Sub-Saharan Africa far outweigh the use of one of the most common methods of data collection in Europe — door-to-door canvassing. The roundtable participants noted that this was perhaps because it is easier in Sub-Saharan Africa to collect data through illegitimate, and at times, illegal, methods; whereas door-to-door canvassing is one of the few ways to gather data legitimately in Europe due to strict data laws. However, the use of personal data to personalise political messages is key to data-driven tactics of influence across Europe, the US and Sub-Saharan Africa. For example, one participant described how a senator in Kenya sent messages to people in his district to share information about the projects he was working on. The people who received these messages were excited to be contacted personally and therefore more motivated to support Uhuru’s Jubilee party. These messages can appear very personal, and are effective in making people feel involved in politics and connected to the political candidate in question. The participants of the roundtable noted that this feeling often overrides any concerns regarding privacy and how the candidate got the phone numbers, limiting the ability to gain public interest in interrogating the topic further.
The issues which are raised around how the data-driven campaigns work, beyond privacy concerns, were often connected to the broader internet system. In the course of the discussion, the group identified four attributes of how the internet works which are specific to the region and impact the use of data-driven campaigning methods:
- Firstly, there is substantially low internet penetration in certain countries, such as only 16% in Sierra Leone, where only 9% of people are on social media. This does not mean that internet-based and data-driven methods do not have influence. People often transmit the information they read online through discussions with family and friends. Furthermore, people with more wealth and access to the internet then also have access to data and the ability to analyse and understand where to target their campaigning efforts, whether online or in person.
- Secondly, the region experiences regular and substantial internet shut-downs. An internet shut-down in Ethiopia disrupted one participant’s attendance in the virtual roundtable. These shut-downs can happen around elections and cause information blackouts which affect the cycle of data-driven campaigns.
- Thirdly, free internet is sometimes provided by Facebook, which can impact how people use social media. One participant stressed the importance of researchers outside the region to account for the reasons people access information the way they do and in particular how engagement with Facebook may be due to the access to internet, rather than any particular feature of the social media platform.
- Finally, data-driven or online techniques are not always the most used methods of influence, in some cases because of the use of traditional methods such as billboards and rallies, in other cases because of the use of illegitimate methods such as buying votes.
Despite the importance of these issues, we found in our roundtable discussion that most funding for research into data-driven elections or internet and politics does not go towards Sub-Saharan African countries. For example, WhatsApp granted one million US dollars to organisations to look at how WhatsApp was being used to spread misinformation. Out of the 20 grants, only one focused on a Sub-Saharan African country, Nigeria.
Digital campaign consultants and firms who work for political parties, often also based in Europe or U.S, carry out their own research in the region as well — but for the purposes of improving their own strategies. One of the well-known methods for this research is A/B testing, in which companies test different messaging, images and channels for communication and record which produce the most responses and feed this back into their strategies. The firms’ research is profit-driven, and aims to answer specific questions about which techniques are “successful”, rather than the questions needed to evaluate these methods in relation to political participation. The firms want to know if they will get more votes, or even just if they will get paid by the political party, but not whether they have caused more conflict or whether information has only reached certain groups and not others. This is partly because the companies rarely stay in the country where they tested these techniques long enough to experience the mid- to long-term outcomes.
Companies rarely stay in the country where they tested these techniques long enough to experience the mid- to long-term outcomes
This testing and research is important as the use of data-driven campaigning methods in the region is developing substantially. As the numbers of people with access to internet rise every year, the way these increasingly pervasive tools are used now will significantly impact how the tactics and methods will develop. Furthermore, while the data-driven practices are already well-established in Kenya, Nigeria and Uganda, many participants questioned how these countries’ practices will influence the uptake of the same tools in Togo or Ghana. Whilst the insights from this roundtable discussion were substantial, they are still only snippets of a much bigger picture of the growing data-driven electioneering ecosystem in Sub-Saharan Africa. More substantive research needs to take place in order for us to have a fuller picture of the changes to political participation that will come as a result. All the participants agreed that more funding needs to go to local researchers in Sub-Saharan Africa, and researchers elsewhere should work closely with local experts. Further, research needs to take into account more voices, focusing not just on the large consultancy firms or political parties but also activist groups and citizen’s experiences. Research should also examine the crossover between those with access to internet and those with only access to information shared offline, such as through conversations or local group meet-ups. Only through interrogative and well-resourced research can there be a better understanding of the specific aspects of the methods which are exported from the US and Europe to Sub-Saharan Africa. It is only then that we will know what impact differences in the region have for these methods, as well as which techniques are tested in the region to be imported back to the companies’ home country and what long-term impacts are created by these short-term experiments in influencing elections.
Amber Macintyre is a researcher and facilitator at Tactical Tech
Thank you to Sasha Ockenden, Christy Lange, and the Data and Politics team for their feedback on this article and to participants for their input into the full report.