The article ‘Systematic discrepancies in the delivery of political ads on Facebook and Instagram’, published in the journal PNAS Nexus, revealed how social media algorithms favour political content sponsored by certain parties for the same amount of budget invested.
The research is a collaboration between researchers from the Politecnico di Milano, the Ludwig Maximilians Universität in Munich and the CENTAI institute in Turin, including Francesco Pierri from the Data Science research group at the Department of Electronics, Information and Bioengineering.
The study analysed over 80,000 political advertisements on Facebook and Instagram before the German federal elections in 2021. The ads were placed by parties across the political spectrum and generated more than 1.1 billion impressions during an election with more than 60 million eligible voters.
Investigating inequalities in online campaigns, significant discrepancies emerged in the effectiveness of advertising and the intensity with which the ads achieved their goals by rewarding the most extremist groups.
The data collected showed that more than 70% of the parties used user profiling in their ads. Furthermore, variations in advertising costs (impressions per euro spent) showed that not all parties achieved the same results with the same budget. The far-right AFD was the most effective with almost six times as many ads as its competitors who had invested the same budget. The Greens were the least cost-effective party.
Another finding was the discrepancies for all parties between targeted and actual audiences. While most parties tended to reach a younger audience than expected, the opposite was true for the extreme right. Pierri and his colleagues hypothesise that the algorithmic bias in ad distribution is based on known voter behaviour.
It is therefore not surprising that targeted political advertising on social media has raised serious doubts among political actors, researchers and society at large. Calls for improved monitoring of this type of advertising to safeguard democratic integrity are growing louder and louder. Public pressure and regulatory efforts (e.g. the Digital Services Act in the EU) have pushed social media platforms to provide public access to political and social ads, allowing researchers to study them on a larger scale.