A new paper published in Political Science Research and Methods has found that using Facebook ads can be a cost-effective survey recruitment tool, particularly in the Global South.
The study used Facebook ads to sample respondents in Mexico, Kenya, and Indonesia, and assess how well these samples performed compared to traditional surveys.
Findings suggest that using this method yields data comparable to that from commercial online samples such as Amazon Turk or Prolific, but at much lower costs, offering a practical and accurate alternative for researchers working in resource-limited settings.
Historically, limited electricity, phone, and internet connectivity in many regions – especially in the Global South – has meant that researchers often relied on in-person survey recruitment, requiring substantial budgets and lengthy fieldwork.
However, today, much more of the world’s population is digitally accessible – in 2022, an estimated 68% of the world’s population had a mobile phone subscription, and approximately 55% had access to mobile internet.
Co-author Dr Francesco Rampazzo, Lecturer in Demography at Oxford’s Department of Sociology and Leverhulme Centre for Demographic Science, explained:
The growth of digital connectivity offers researchers a chance to reach diverse participant pools and conduct studies in areas where in-person contact is difficult.
This connectivity is especially valuable for researchers with limited budgets or those working in settings where traditional, resource-intensive research methods are impractical, such as during natural disasters, violent conflicts, or pandemics.
"We began this research project before the COVID-19 pandemic," added co-author Dr Leah Rosenzweig, Senior Fellow at the University of Chicago’s Center for Global Development. "But the pandemic demonstrated the value of online recruitment methods in a way we hadn’t fully anticipated. This approach allows researchers to gather valuable data from populations that were previously expensive to reach, at a fraction of the cost of traditional surveys."
Estimated to have 3 billion users as of 2023, Facebook is the most widely used social media platform in the world. Given this massive user base, the platform offers researchers potential access to nationally, culturally, and demographically diverse global populations.
Facebook can be used to quickly and cheaply recruit respondents, with surveys taking between 1-3 weeks to complete, and costing an average of US $1.03 per completed response.
The findings from this study confirm that Facebook ads can recruit survey samples quickly and affordably, producing comparable data to a commercial online sample for a fraction of the cost.
However, participants sourced from Facebook data tend to be more educated than the general population. To combat this, the paper recommends the use of survey weights to counter representational biases.
The authors caution that using Facebook data will not be suitable in all contexts. The platform will be most successful in regions where phone and internet penetration is widespread, literacy rates are high, and recent census population data is available to allow weighting.
Researchers should also consider users’ socio-political context, particularly in authoritarian states, where survey responses may be affected by concerns about government surveillance.
Dr Francesco Rampazzo added:
Facebook should not completely replace gold-standard, resource-intensive in-person field surveys for recruiting nationally representative samples in the Global South.
Our Facebook samples tend to be more highly educated and slightly younger than the national populations in our case countries, which slightly skews descriptive inferences on political engagement and public policy towards more educated perspectives.
However, Facebook represents a valuable tool that, when used well, could open new frontiers in public opinion research.
The study emerged from a collaboration formed during Duke University’s Summer Institute in Computational Social Science. Co-author Dr Katherine Hoffman Pham said:
We began this work after observing that Facebook surveys were increasingly being used by researchers in diverse settings around the world, but little was known about their representativeness outside of a handful of contexts.
We find that Facebook surveys are promising but flawed – in general, they enable quick and cheap access to populations in the Global South, but the costs of targeting some hard-to-reach population segments is prohibitively high (if these populations can be reached at all).
We demonstrate a principled approach to considering sources of representation error in Facebook surveys, and discuss some of the challenges, nuances, and practical strategies for using this platform. We hope to provide helpful guidance to researchers interested in using similar methods.
The full paper can be read here.