Facebook political advertisement analysis

Observers have speculated that political ads on social media contribute to growing divisions in American society, engaging in partisan mudslinging and uncivil attacks over respectful discussion of the issues.

But no one so far has examined a complete universe of political ads from a major social media platform. We combine decades of political science research with the latest advances in machine learning to analyze thousands of political ads on Facebook.

Social media ads are a growing feature of modern political campaigns. More and more, political candidates opt for connecting with voters online, taking advantage of social media platforms’ advertising tools to identify and reach thousands of supporters easily, instantly, and cheaply. In 2018, US political campaigns spent $623 million on social media ads. In 2020 — during a contentious battle for the presidency amidst a once-in-a-lifetime pandemic — this spending more than doubled to $1.34 billion.

For a long time, we knew little about the content and targets of digital political ads. Digital ads are hard to study. They disappear once a new webpage loads. They are microtargeted, meaning only specific individuals meeting specific criteria ever see them. And there is no FEC or FCC regulation in the US requiring that digital political ads be recorded.

But in the last two years, major technology companies have begun creating archives of political ads. In May 2018, Facebook established the Facebook Ad Library – a centralized database of all political ads run on its platform. The Ad Library provides both a snapshot of the ad and important information about the ad: how much the campaign spent on it, how many impressions it received, and who was in its audience.

Scraping this data with the Ad Library’s API , we investigate the content of political ads run by US Congressional candidates on Facebook. We engage in topic modeling to identify the socioeconomic issues that candidates mention most often in political ads. We engage in sentiment analysis to examine the tone – positive or negative – of political ads. And we pair our findings with information about the candidate who ran the ad and the voters who saw the ad in order to better understand the dynamics of campaign communication in the modern era.

Learn more

According to one source, the total spending on ads for the 2020 US Presidential elections is estimated $6.89 Billion, which is nearly 60% more than what was spend for the 2016 election campaings. A major portion of this is spent on TV ads. However, with exponential increase in social media users, around $1.34 Billion is estimated to be spent on Digital advertising, which is a staggering 200% jump as compared to 2016 election campaing.

The next steps for us would be analyse the ads data pertaining to 2020 US Presidential Elections. It is particularly expected to be an interesting study given all that had happened in 2020.