Advertisers Jump on Coronavirus Bandwagon: Politics, News, and Business

Yelena Mejova, Kyriaki KalimeriarXiv 2020 · arXiv

TL;DR

During the early outbreak of COVID-19, diverse actors quickly co-opted the crisis to push political, commercial, and philanthropic agendas through paid Facebook advertisements. This study reveals that official public health messaging was heavily drowned out by fragmented alternative narratives, with approximately 5.3% of unique ads containing clear instances of misinformation or conspiracy theories.

What Problem Does This Paper Address?

When a public health crisis strikes, the World Health Organization (WHO) warns not just of a pandemic, but of an infodemic—an overabundance of information, both accurate and false, that makes it hard for people to find trustworthy guidance. While researchers frequently study organic social media posts to track rumors, paid advertising remains a critical blind spot.

Advertisers can bypass organic discovery algorithms to micro-target specific demographics. This paper addresses how non-public health actors (politicians, businesses, news media, and non-profits) leveraged the early days of the coronavirus outbreak on Facebook to advance their own agendas, potentially undermining official, fact-based public health communication.

Methods

The authors collected and analyzed paid advertisements using the following pipeline:

  1. API Querying: They queried the Facebook Ads Library API on February 20, 2020, using the keywords "coronavirus" and "covid-19" to capture both active and completed campaigns.
  2. Open Coding & Classification: The authors manually categorized 78 distinct Facebook Pages into six sectors: public health, non-profit, news, political, business, and personal.
  3. Network Analysis: To see if different pages shared similar messaging, they computed the Jaccard similarity between the text of all ads from each page, building a co-occurrence network visualized via Gephi's ForceAtlas2 layout.
  4. Targeting & Emotion Analysis: They aggregated demographic (age/gender) and geographic targeting data. They also applied the DepecheMood++ lexicon to evaluate fine-grained emotional connotations (e.g., fear, amusement, inspiration) across ad text.
  5. Misinformation Annotation: Ads were manually cross-referenced with fact-checking platforms like Snopes and Wikipedia to flag potentially erroneous or malicious claims.

Datasets

The study introduces a novel dataset consisting of 923 ads from 34 countries. Because the volume was heavily concentrated, the analysis primarily focuses on the 359 ads originating from the United States, followed by Italy (228) and India (64).

The dataset is unique because it includes backend metrics provided by the Facebook Ads Library, such as estimated financial expenditures, total impressions, and demographic/regional targeting distributions.

Key Insights and Ideas

  • The Public Health Deficit: Official public health organizations contributed a minuscule fraction (only 6%) of the total advertisers in the dataset. Paid spaces were dominated by news media and political actors.
  • Partisan Weaponization: Coinciding with the 2020 US Democratic primary season, the virus was immediately weaponized. Ads from both sides of the political spectrum used the epidemic to either attack or defend political figures, distracting from actionable health advice.
  • Commercial Exploitation: Unverified commercial entities quickly filled the void, advertising "Anti-Virus Masks" (lacking medical certification) or leveraging market panic to sell financial and stock advice.
  • Demographic Targeting of Misinformation: Misinformation wasn't distributed evenly. Political ads containing potential misinformation heavily targeted older demographics (65+), who are simultaneously the most vulnerable to the virus and highly active in political discourse.

Key Results

  • Volume Trends: The volume of ads spiked dramatically around February 5, 2020, corresponding with the first confirmed case in Wisconsin.
Figure 1: Number of active ads per day for countries hav- ing at least 10 ads in the dataset.
  • Prevalence of Misinformation: Out of 359 US ads, 16 contained clear misinformation (representing 5.3% of unique ad texts). These ranged from egregious bioweapons conspiracy theories to unverified claims that warm weather would kill the virus.
  • Top Spenders: The highest expenditures came from the American Medical Association, National Nurses United, the Melinda Gates Foundation, and Michael Bloomberg’s presidential campaign ("Team Mike 2020").
Figure 4: Top pages by expenditures (left) and impressions (center) and dollars spent per impression (right).
  • Emotional Framing: "Fear" (afraid) was the most prominent emotion detected across news and non-profit ads. Interestingly, "inspired" was the second highest, driven largely by non-profit donation solicitations.

Limitations and Future Work

The primary limitation stems from the Facebook Ads Library's selection criteria. The library is designed to expose ads that Facebook deems relevant to "social issues, elections or politics." Consequently, standard, non-political public health campaigns may have been omitted by Facebook's filtering mechanisms, artificially depressing the observed share of public health ads.

Additionally, the data represents a snapshot from the very early stages of the pandemic (up to February 20, 2020). The authors note that as the crisis evolved, narratives likely shifted. They call for closer collaboration between social media platforms and public health researchers to establish better quality control and real-time monitoring of paid health communication.

Summary updated

Abstract

In the age of social media, disasters and epidemics usher not only a devastation and affliction in the physical world, but also prompt a deluge of information, opinions, prognoses and advice to billions of internet users. The coronavirus epidemic of 2019-2020, or COVID-19, is no exception, with the World Health Organization warning of a possible "infodemic" of fake news. In this study, we examine the alternative narratives around the coronavirus outbreak through advertisements promoted on Facebook, the largest social media platform in the US. Using the new Facebook Ads Library, we discover advertisers from public health and non-profit sectors, alongside those from news media, politics, and business, incorporating coronavirus into their messaging and agenda. We find the virus used in political attacks, donation solicitations, business promotion, stock market advice, and animal rights campaigning. Among these, we find several instances of possible misinformation, ranging from bioweapons conspiracy theories to unverifiable claims by politicians. As we make the dataset available to the community, we hope the advertising domain will become an important part of quality control for public health communication and public discourse in general.
BibTeX
@misc{mejova2020advertisers,
  author = {Yelena Mejova and Kyriaki Kalimeri},
  title = {{Advertisers Jump on Coronavirus Bandwagon: Politics, News, and Business}},
  year = {2020},
  eprint = {2003.00923},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2003.00923},
}
Figure