Gender-based discrepancies in the algorithmic delivery of political ads on social media
This study analyzes over 110,000 political ads from the 2024 European Parliament elections and reveals that algorithmic delivery on social media systematically exposes men to a significantly higher share of populist and far-right advertising than women, thereby distorting voter exposure and threatening electoral integrity.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine the internet as a massive, bustling town square where political parties set up booths to hand out flyers (ads) to voters. In this town square, there isn't a human handing out the flyers; instead, there is a giant, invisible, and somewhat mysterious robot (the algorithm) that decides who gets which flyer.
This paper is a report card on how that robot behaved during the 2024 European Parliament elections. The researchers wanted to see if the robot was playing favorites based on gender, specifically when handing out flyers from "populist" parties (groups that often claim to speak for "the common people" against "the elite").
Here is the breakdown of their findings in plain English:
The Big Discovery: The Robot Has a "Male Bias"
The researchers looked at over 110,000 political ads that reached more than 7 billion people across 25 European countries. They found a clear pattern: The robot was much more likely to show ads from populist parties to men than to women.
Even when the parties didn't explicitly tell the robot, "Only show this to men," the robot still did it anyway. On average, ads from populist parties reached a male audience that was 6 percentage points higher than what you would expect if the distribution were perfectly fair.
The "Populist" vs. "Normal" Party Test
To understand if this was just a general quirk of the robot, the researchers compared two types of flyers:
- Populist Party Flyers: These are often emotional, divisive, or anti-establishment.
- Non-Populist Party Flyers: These are more traditional or centrist.
The Result: The robot treated them differently.
- Populist ads were like magnets for men. They ended up in the hands of a much larger male crowd.
- Non-populist ads were distributed much more evenly between men and women.
It's as if the robot has a "gut feeling" that men are more interested in populist messages, so it starts handing those specific flyers to men more aggressively, even if the person writing the flyer didn't ask for that.
The "Far-Right" vs. "Far-Left" Twist
The researchers also looked at the political spectrum, like a line from left to right.
- Far-Right Parties: Their ads were heavily skewed toward men (similar to the populist finding).
- Far-Left Parties: Interestingly, their ads were actually skewed toward women.
- Centrist Parties: These were the most balanced, reaching men and women fairly equally.
This suggests the robot isn't just "male-biased" in general; it has learned specific associations. It seems to have learned that "Far-Right = Men" and "Far-Left = Women" based on past behavior, and it reinforces these stereotypes automatically.
Why Did This Happen? (The "Echo Chamber" Effect)
The paper suggests the robot isn't necessarily "evil" or trying to be unfair. Instead, it's acting like a predictive autopilot.
Imagine the robot looks at history and sees that, in the past, when a Far-Right party posted a flyer, men clicked on it more often than women. The robot thinks, "Aha! Men like this stuff. I'll show it to more men to get more clicks."
This creates a feedback loop:
- The robot shows the ad to more men.
- More men see it and click.
- The robot sees the clicks and thinks, "See? I was right! Show it to even more men."
- Women never get to see the ad, so they can't click, and the robot never learns that women might be interested too.
The Problem with the "Black Box"
The biggest issue the paper highlights is that the robot is a black box. The companies that own the social media platforms (like Meta/Facebook) don't fully explain how the robot makes these decisions.
The researchers tried to account for everything they could:
- Did the ad have a picture of a man? (No, they checked the content).
- Did the party specifically ask to target men? (No, they filtered out those ads).
- Was it a different day of the week? (They checked the timing).
Even after removing all those obvious reasons, the gender gap remained. This means the bias is hidden deep inside the robot's code, likely driven by the complex way it learns from past user behavior.
The Bottom Line
The paper concludes that this isn't just a small glitch; it's a structural problem. If the robot keeps handing Far-Right flyers mostly to men and Far-Left flyers mostly to women, it prevents voters from seeing the full range of political ideas. It narrows the conversation and might make political divides even deeper.
The authors argue that we need to audit the robot. Just like we check election results for fraud, we need to check these algorithms to make sure they aren't secretly rigging who gets to hear what political message. Without this check, the "town square" isn't actually open to everyone equally.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.