August 27, 2025 | News | Recommended Reading
Facebook Ads: A Tool for Scientific Data Collection?
Study examined how Facebook ads can be used to recruit survey participants
Researchers at the Max Planck Institute for Demographic Research (MPIDR) and Bielefeld University investigated how ad design influences the quality of survey responses on Facebook. They tested various ad designs and evaluated the results using six quality markers. Their findings reveal that ads explicitly referencing specific topics generate more clicks, yet often result in lower-quality responses. Neutral ads, on the other hand, lead to higher-quality responses.
Researchers at the Max Planck Institute for Demographic Research (MPIDR) and Bielefeld University analyzed how ad design influences the quality of survey responses on Facebook. © iStockphoto.com / Boy Wirat
In demography and the social sciences, survey data play a crucial role in answering questions about key social issues, such as social inequality, political attitudes, and social cohesion. Against the backdrop of declining participation rates in traditional survey methods, such as telephone or in-person interviews, alternative formats, such as web-based self-administered surveys, are becoming increasingly important. These online surveys often use social media advertisements to recruit participants.
At the Max Planck Institute for Demographic Research (MPIDR)'s Department of Digital and Computational Demography, researchers investigate how digital trace and social media data can be used and evaluated. In a recent study, Jessica Donzowa (MPIDR), Simon Kühne (University of Bielefeld), and Zaza Zindel (University of Bielefeld and the German Center for Integration and Migration Research), investigated how the design of Facebook ads affects the quality of survey responses.
The research team tested five different Facebook ad designs to recruit survey participants. The ads either referred to climate change or immigration, or were neutral. The researchers evaluated the results using six quality markers:
- Survey abandoned
- Completed hastily
- Inconsistent answers
- Questions not answered
- Attention check passed or failed
- Provision of an email address for further surveys
© MPIDR/ iStockphoto.com / Halfpoint (Demo Kids) / iStockphoto.com / Brad Pict (LanguageBubbles/Sprechblasen) / iStockphoto.com / Nicole Effinger (Blatt mit Herz/Leaf with a heart)
“These indicators allow us to better assess how carefully participants completed the survey. We used logistic regression to investigate the relationship between the ad design and these quality indicators. We also included differences in age, gender, income, and whether people used a smartphone or computer in the analysis," Donzowa explains.
The researchers found that ads explicitly referencing immigration or climate change generated more clicks and were more cost-effective than neutral designs. However, these responses were often of lower quality. Participants skipped questions more often or gave inconsistent answers. "In contrast, we achieved better results with neutral ads. Participants generally gave higher-quality answers here. Neutral designs are more suitable for general research surveys," Zaza Zindel explains.
The researchers point out, however, that their findings may not be transferable to other platforms, such as Instagram, TikTok, or LinkedIn, as they have different user and interaction patterns. "A bigger challenge for all studies that use social media ads to recruit survey participants is the platform's opaque algorithm, which can also change over time," says Donzowa. Research methods must be regularly reevaluated to remain reliable.
This study provides guidance for researchers who want to use Facebook ads to recruit survey participants. The choice of ad design should depend on the survey's objectives and the country's context.
© MPIDR
Original Publication
Jessica Donzowa, Simon Kühne, Zaza Zindel: From Clicks to Quality: Assessing Advertisement Design’s Impact on Social Media Survey Response Quality in mda (2025); DOI: 10.12758/mda.2025.05