Ads and Voter Behavior: Lessons from 2020 for 2024
Analyzing the impact of past ad campaigns on future voter decisions.
Xinran Miao, Jiwei Zhao, Hyunseung Kang
― 6 min read
Table of Contents
In every election, candidates try to win over voters with ads. These ads can be positive, highlighting the candidates' qualities, or negative, pointing out their opponents' flaws. In recent years, campaign teams have tried to see how effective these ads really are. This article looks back at the 2020 U.S. presidential election and wonders if we can use what we learned to inform the ads for the 2024 elections.
Why We Care About Ads
Ads can shape how people think about candidates. They can influence whether someone decides to vote and for whom. After the 2020 election, a big question came up: if we ran a negative ad campaign against Trump in 2024, would it change how people in Pennsylvania (a very important state) vote?
To answer this, we could conduct a random experiment where a group of voters sees one type of ad while another group sees something completely different. However, these experiments can be very costly. Instead, we're looking for a cheaper and faster way to make predictions based on past data.
The 2020 Experience
Back in 2020, researchers tested negative ads against Donald Trump. They found that these ads didn’t really change Voter Turnout in Pennsylvania. Given the tension and unique circumstances of that election, it raises the question: will the same ads work in 2024? The landscape has changed. New issues like women's rights, inflation, and international conflicts are now on voters’ minds.
Given these changes, we want to figure out if the past results still apply.
The Cost of Experiments
Running a new experiment for 2024 could cost millions, while a campaign in 2020 spent nearly $9 million on ads. Some organizations are already gearing up with hefty budgets, one reportedly went as high as $450 million to test different ads for effectiveness. So, what’s the alternative?
We propose using knowledge from 2020 to evaluate potential outcomes for 2024 ads. By comparing the voter Demographics and context of both elections, we can try to make a reasonable guess about how effective certain ads will be.
Our Approach
Our approach includes backup plans for understanding how differences between the two elections could influence the outcome. We use something called a sensitivity analysis to measure these unobservable differences.
In simple terms, we’re trying to estimate how different the two elections are and adjust our results accordingly. Our goal is to see if the way voters reacted to ads in 2020 can help predict how they'll react in 2024.
Estimating Ad Impact
We focus on the effect of running a negative ad against Trump on Pennsylvania voters for the upcoming election. Pennsylvania is critical because it has a large number of electoral votes. By looking at voting patterns in various counties, we can get a better picture of where the ads might work and where they might not.
Using different methods, we break down the expected outcome by different groups of voters. For example, are men or women more likely to be swayed by negative ads? What about urban versus rural voters?
Key Findings
Our analysis shows mixed results. In Fulton County, which heavily supported Trump in 2020, negative ads might slightly reduce voter turnout. However, in many other counties, the ads aren't expected to make a significant impact.
In terms of voter subgroups, we found that negative ads could trigger decreased turnout among women living in rural areas with less education, while they might boost turnout among more educated, non-female voters living in urban areas.
Exploring Related Research
Our methods are not new but build on previous research on how to generalize treatment effects from one group to another. For this analysis, we used data from 2020 and 2024 to make predictions while accounting for the differences in voter demographics.
The goal is to ensure that we are accurately assessing the impact of negative ads without having to run expensive new experiments. By using past data smartly, we can still get useful insights.
Sensitivity Analysis Explained
Whenever we analyze data, we have to consider the possibility that unseen factors could skew our results. That’s where sensitivity analysis comes into play. It helps us gauge how much unknown changes between the two elections could alter our conclusions.
If we find that certain assumptions hold, we can be more confident in our predictions. If not, we need to be cautious in our interpretations.
Data Collection Basics
Gathering the right data is crucial. In our case, we collected demographic information such as age, gender, and party affiliation from registered voters in Pennsylvania. This way, we can see how different groups might react to the ads.
To make sure our data is reliable, we carefully recoded the information to fit our needs. This step ensures that we’re using consistent definitions across both elections.
Moving Forward with Predictions
Using our approach, we presented estimations about the effectiveness of negative ad campaigns in Pennsylvania for 2024. We broke this down by counties and voter demographics, painting a clearer picture of how voters are likely to respond.
Interestingly, the patterns suggest that counties that largely supported Trump in 2020 might see negative ads as unconvincing, while areas that leaned Democratic could respond better.
The Importance of Context
The political landscape in 2024 is different from 2020. Issues like inflation and women's rights are now front and center. These factors may create new sensitivities among voters that the previous campaigns didn’t account for.
Understanding these new dynamics is vital for any future ad campaigns. After all, ads that worked before may not have the same effect later due to the shifting concerns among voters.
Conclusion
Ultimately, while running a fresh randomized experiment is the gold standard for understanding ad effectiveness, it’s not always practical or affordable. By using transfer learning from past elections, we can glean valuable insights into how digital ads might impact voter turnout in the future.
Our analysis highlights the need for a thoughtful approach to understanding voter behavior, especially as new issues come to the forefront.
Continuous Evaluation
As we approach the 2024 elections, continuous evaluation of voter sentiments and responsiveness to ads will be essential. By learning from past campaigns and adapting accordingly, campaign teams can better connect with voters and maximize their impact.
Understanding the subtleties of voter demographics and their evolving concerns will be central to winning their support.
So, buckle up for the 2024 election, folks! It’s going to be an interesting ride full of ads, debates, and perhaps a bit more drama than necessary!
Title: Transfer Learning Between U.S. Presidential Elections: How Should We Learn From A 2020 Ad Campaign To Inform 2024 Ad Campaigns?
Abstract: For the 2024 U.S. presidential election, would negative, digital ads against Donald Trump impact voter turnout in Pennsylvania (PA), a key "tipping point" state? The gold standard to address this question, a randomized experiment where voters get randomized to different ads, yields unbiased estimates of the ad effect, but is very expensive. Instead, we propose a less-than-ideal, but significantly cheaper and likely faster framework based on transfer learning, where we transfer knowledge from a past ad experiment in 2020 to evaluate ads for 2024. A key component of our framework is a sensitivity analysis that quantifies the unobservable differences between past and future elections, which can be calibrated in a data-driven manner. We propose two estimators of the 2024 ad effect: a simple regression estimator with bootstrap, which we recommend for practitioners in this field, and an estimator based on the efficient influence function for broader applications. Using our framework, we estimate the effect of running a negative, digital ad campaign against Trump on voter turnout in PA for the 2024 election. Our findings indicate effect heterogeneity across counties of PA and among important subgroups stratified by gender, urbanicity, and education attainment.
Authors: Xinran Miao, Jiwei Zhao, Hyunseung Kang
Last Update: 2024-11-01 00:00:00
Language: English
Source URL: https://arxiv.org/abs/2411.01100
Source PDF: https://arxiv.org/pdf/2411.01100
Licence: https://creativecommons.org/licenses/by/4.0/
Changes: This summary was created with assistance from AI and may have inaccuracies. For accurate information, please refer to the original source documents linked here.
Thank you to arxiv for use of its open access interoperability.