Political microtargeting (PMT) is a relatively new technique that uses citizens' personal data to create tailored messages. For example, a nurse will be shown an advertisement in which a political party promises to ensure good healthcare, whereas a teacher will receive messages on matters relating to education.

Microtargeting makes messages more relevant to the recipient and thus increases the likelihood of winning them over. UvA communication scientist Tom Dobber investigated PMT in the Netherlands. He concludes that political parties in the Netherlands also use PMT and that it can be effective, but whether it forms a potential threat to democracy depends primarily on the players who are using it.

Brexit and Donald Trump

Two well-known examples of microtargeting being presented in a particularly negative light in the news were the surprisingly successful Leave campaign in the UK, leading to Brexit, and the unexpected election of Donald Trump as president of the United States. In both cases, political microtargeting was quickly identified as the cause.

These two events led UvA communication scientist Tom Dobber to examine the role and effects of PMT in the Netherlands. ‘We barely knew anything about PMT and mainly regarded it as a very American phenomenon. However, does PMT also play a role in the Netherlands where political parties have much smaller campaign budgets, there are different privacy laws and the electoral and party systems are also very different?’ Dobber wondered.

How does political microtargeting work?

Political microtargeting deploys sophisticated techniques to use citizens' personal data to draw up personal profiles and tailor messages accordingly. This is the kind of data that you might add to an online profile yourself, but could be also location data (the type of places you visit) and payment data (the type of products you purchase), and naturally also includes your online surfing behaviour.

‘At an individual level this data might not say all that much, but on a large scale it enables you to identify behavioural patterns and determine which profiles will be sensitive to which messages. Anyone can buy this data. You can already do a lot with just a few hundred euros,’ says Dobber.

Political microtargeting in the Netherlands

Dobber examined the extent to which Dutch political parties use PMT, what Dutch people know about PMT and how they feel about it, the possible effects of PMT and the influence of PMT on the effects of deep fakes (fake videos in which someone says or does something they have never actually said or done in real life). He held interviews with political parties, conducted a panel study among the Dutch population, carried out a field experiment during municipal elections in Utrecht and finally conducted an online deep-fake experiment.

Dobber concludes that:

• Political parties in the Netherlands also use microtargeting. However, he identified a modified form that stems from the different context (campaign budgets, legislation and political system).
• The Dutch population regard PMT in quite a negative light and have numerous concerns about privacy. ‘And the more people worry about privacy, the more negatively they regard PMT,’ Dobber discovered. This downward spiral is worrying if it causes people to lose confidence in democracy and turn away from it.
• Although participants in the field experiments did appear to think more positively about the political party in question after receiving PMT messages, they were no more likely to vote for this party as a result. ‘So it does seem to work, but doesn’t have a huge effect,’ concludes Dobber, ‘although this was of course a relatively small-scale experiment.’
• Microtargeting techniques may further enhance the negative effects of deep fakes.

Is political microtargeting a threat?

Does political microtargeting form a threat to our elections? ‘The answer is relatively nuanced,’ says Dobber. ‘People have always tried to influence citizens, and this isn’t such a bad thing in itself. For example, tailored messages also allow you to share information with citizens that they truly find relevant. As a result, you can motivate citizens who have not previously been politically active to participate in the political process and vote. This can be positive, as long as it is carried out by legitimate players who are acting fairly and in good faith,’ says Dobber.

‘However, influence by PMT becomes unacceptable when it turns into manipulation, if it abuses people's vulnerabilities by responding primarily to their fears, if it resorts to lies or unlawfully obtained personal data and if it is carried out by non-legitimate players,’ concludes Dobber.

‘By way of reassurance, elections are complex events and these kinds of complex events can be influenced, but getting an election to move in the desired direction and in a controlled manner is fortunately more difficult.’

The future of political microtargeting

‘As technology advances, we will be seeing smarter techniques that can even target people by their personality, among other things by analysing their use of language. Specific messages could then be created for introverts versus more extrovert profiles, for example,’ says Dobber. ‘As a citizen, you would no longer realise that this was a targeted message.’

Dobber would like the use of extensive targeting to be limited to legitimate players who are acting in good faith. For his part, he will now focus on the role and effects of deep fakes and disinformation and, in the run-up to the American elections, will analyse the effects of overt and covert election advertisements.