Project 2

Thesis

This study aims to evaluate the accuracy of polling data in the 2024 U.S. presidential election by examining potential biases in favor of specific candidates. The focus will be on identifying whether polling organizations systematically overestimate support for Kamala Harris compared to Donald Trump. By using previous election data to reveal any potential trends in current data, my goal is to predict outcomes of the 2024 election.

Model Analysis

The first table here shows the accuracy of Donald Trump’s predictions for 2020, with an astounding .98 R2 value. Joe Biden had a .97 R2 value, which is also quite high. However, using the same parameters, Trump’s estimates were closer than Biden’s for 2020 by a slim margin.

One aspect worth mentioning is the scale of Biden’s prediction. Washington DC received over 90% democratic votes, but it is not far off from the line of best fit for this graph.

Next, let’s take a look at some predictions.

Predictions:

Prediction Analysis

Overall, these predictions seem to be pretty close to accurate. However, it would prove my thesis to be incorrect in thinking that polls may be a tad left-leaning.

Harris has a considerable lead in:

  • Massachusetts

and Trump has considerable leads in:

  • Montana

  • Nebraska

  • Texas

Trump also runs a higher average vote per state by a small margin. The averages are not far from each other, but where the averages really stand out is in the swing states. These are states that can make or break the election for either candidate and can come down to the flip of a coin on election day. States like my home state of Pennsylvania, Georgia, North Carolina, and Arizona are some of the big-named swing states.

Resources

https://projects.fivethirtyeight.com/polls/

  • Polling Data (I used president_polls.csv, president_polls_historical.csv, and 2020_election_results.csv (attached in submission))

ChatGPT

  • I asked ChatGPT for pivot_wider and longer examples, prediction graphing models, and help making some of my plots prettier.

https://profgarrett.github.io/course_eda/r00-index.html

  • Course notes helped me mutate data to fit my needs and report formatting.