Using ‘Ethically Sourced’ Social Media Data to Study Health Marketing Disparities
Authors
Sherry Emery
Director, Social Data Collaboratory
January 2025
Social media users’ donated data reveals how targeted advertising may promote health disparities.
As a newly minted health economist in the late nineties, I landed a position working with an amazing tobacco control researcher in San Diego who hired me to work on the quantitative side of his research. That work led me to a trove of data that would drive the next decade of my research.
It was a fortuitous time to stumble into tobacco research. Forty-one states had won lawsuits against the tobacco industry and were using their settlement money to pay for advertising campaigns to curb tobacco use. States took different approaches to their campaigns, providing a natural experiment that allowed my colleagues and me to assess the impact of various tobacco control campaigns. We gained valuable insights into what works and what does not in anti-tobacco public health messaging and how to undo the effects of the tobacco industry’s targeted advertising.
“Social media companies have strong incentives to keep user data from outside researchers despite the potential good that could come from understanding how social media marketing and algorithms affect the public.”
Around the same time, social media was beginning to become popular, and it seemed important to try to capture whatever tobacco-related information was circulating on social media, too. In a moment of hubris, I wondered, “How hard could it be to collect and analyze social media data?”
It turns out that it is really hard. It also turns out that there was almost no anti-tobacco messaging on social media, but there was (and continues to be) a ton of tobacco marketing.
As my team developed new methods to study tobacco marketing on social media and how it affects tobacco use, I quickly realized that there was great potential to ask broader questions about how social media influences our opinions and behaviors surrounding everything from vaccines to cannabis use to voting.
We’re using “donated data” to see what social media users see.
Since then, I’ve been on a mission to figure out how to get social media data. At first, social media platforms provided access to a lot of data. But when Cambridge Analytica was accused of improperly accessing and misusing Facebook users’ data, Facebook cracked down on access.
As the platforms started limiting access or shutting it down entirely, I began reading about what researchers in Europe had started trying—getting regular people to send them data from their own social media accounts. This idea—data donation—is possible because the terms of services for social media providers allow the companies to collect personal data about us. But we the users still own that data and can request it from the platforms. I thought, “What if we just asked social media users to share their data with us?”
To test the idea, we asked 1,100 participants in NORC’s AmeriSpeak panel if they would send us their Facebook and Instagram data. Roughly 800 of them had accounts on the platforms; half agreed to participate, and about 200 followed through by requesting their data and sharing it with us.
We’ve just started sifting through the users’ data, including over 7,000 ad views and 14,000 videos, as well as the pages they visited on the apps, groups the users signed up for, and off-Facebook web searches and website visits. Similarly, we got users’ Instagram data, including nearly 10,000 ads viewed and over 26,000 videos watched.
One of the great things about our approach is that because we asked for user social media data as part of an AmeriSpeak survey, we can match the panel participants’ social media data with their demographic information, such as age, gender, race, and income, which they shared with AmeriSpeak when they became panelists, which we were unable to do in past research.
For the first time, we were able to connect what people are doing on social media with their demographics. In the past, you always had to make some inferences because the platforms did not provide that information. With the donated data, we can see that different groups are getting different types of advertising, and it’s immediately obvious when you see the graphs that there are some important equity implications.
“With donated data, we can see that different groups are getting different types of advertising, and it’s obvious there are some important equity implications.”
Our findings reveal potential equity issues in paid advertisements.
We haven’t yet had the time to dig into all the data. But a quick and dirty analysis of the types of paid advertisements people see identified a couple of potential equity issues. Black respondents were 50 percent more likely to get food advertising than white respondents, and Latinos were twice as likely to be served food ads as their white counterparts. Researchers already know that food deserts contribute to inequities in healthy food access.
The brands, the social media platforms, and algorithms targeting ads to select groups may be another important driver of food-related health inequities. By contrast, about 16 percent of white respondents were served ads about online or home school curricula compared to 4 percent of Black respondents and 5 percent of Latino respondents.
But this analysis just scratches the surface of marketing campaigns’ influence on social media users. Paid advertising is a small fraction of social media product promotion. My current obsession is the nontraditional ways brands market their products on social media. Brands put resources into cultivating influencers and developing engagement strategies that incentivize consumers to act as brand ambassadors themselves. By doing this, brands propagate their marketing without paying for traditional advertising.
Some brands like Taco Bell are also brilliant meme artists and will get consumers to repost memes they think are funny and tie into Taco Bell products. Taco Bell also has content-sharing partnerships with brands like Mountain Dew and Weedmaps, an app that helps people locate cannabis products and dispensaries. Lots of brands have similar strategies—I’m just currently obsessed with fast food promotion.
We are exploring how we can collect donated data from an array of other social media platforms like TikTok or Snapchat, as well as from more traditional data sources like fitness trackers or insurance claims. We may also be able to connect social media data with data from electronic medical records to assess the health impacts of social media marketing.
We are working to build out our donated social media data collection and analysis capability and offer this unique tool to our partners or inform our projects at NORC. It could be helpful to study people’s receptivity to public health messages and other types of public health research.
It may also augment VoteCast, a partnership between NORC and The Associated Press that measures voter sentiment, and other election-related research. The AP-NORC Center’s efforts to answer questions about how media influences public opinion may also benefit from social media data access.
Social media users see donated data as “ethically sourced.”
Social media companies have strong incentives to keep user data from outside researchers despite the potential good that could come from understanding how social media marketing and algorithms affect the public. Policymakers need good research to regulate social media, so the companies have no incentive to make that easy.
However, social media users do have incentives to share their data with researchers. We did focus groups with the AmeriSpeak respondents, and they appreciated the opportunity to participate. Some called the way we collected donated social media data with their consent “ethically sourced data.” They appreciated being able to give their data to researchers, knowing they would use it for projects related to health and public policy. It gave them control over what was happening to their data, something they do not feel they have had with the social media platforms.
Suggested Citation
Emery, S. (2025, January 10). Using ‘Ethically Sourced’ Social Media Data to Study Health Marketing Disparities [Web blog post]. NORC at the University of Chicago. Retrieved from https://www.norc.org.
