Voting is the cornerstone to America's democracy. Voter turnout has the ability to drastically change the course of history. In the 2020 election 154.8 million people voted. Although this was a record number of people, it reflected only 62.8 percent of Americas voting age population. Citizens lack of participation in elections can allow the US government to run unchecked.
Understanding the reasons why people don't vote can help shine a light on where there is misinformation or a lack of understanding on the significance of voting and where the current system of voting has flaws. By identifying individuals attitudes on the success of the US government and their personal experiences and attitudes about voting, we can pinpoint what values and attitudes are associated to lack of voting. The goal of this project is to identify and use a relationship between voters opinion on certain aspects of government and the voting system compared to whether or not someone will vote.
If successful, this work may yield a classifier which predicts the likelihood that an individual will vote based on their opinions. By identifying this relationship and what contributes to this relationship, we can identify where change needs to be made in communicating information about voting and how to pursuade people to vote.
We will use a FiveThirtyEight dataset that contains a file of all the questions given to participants as well as a csv file that contains their responses. Not all of the questions will be used. Instead we will focus on questions pertaining to:
| Question number | data type | data format | description |
|---|---|---|---|
| Q1 | int | int 1 or 2 | asks participants if they are US citizens |
| Q2_1 through Q2_5 | int | int 1 through 4 | asks participants to rank from not important to very important certain things are for being a good American |
| Q4_1 through Q4_6 | int | int 1 through 4 | asks participants to rank on a scale from a scale of no impact at all to significant impact how much certain parts of the government have an impact on their life |
| Q5 | int | int 1 or 2 | Asks participants whether or not it really matters who wins 2020 election |
| Q7 | int | int 1 or 2 | Asks participants if they believe changes are needed in the structure of the government |
| Q16 | int | int 1 through 4 | Asks participants how easy they think it is to vote in elections |
| Q18_1 through Q18_10 | int | int 1 through 2 | Asks participants if they have ever confronted certain restrictions when they have tried to vote |
| Q20 | int | int 1 through 2 | Asks participants if they are currently registered to vote |
| Q26 | int | int 1 through 4 | Asks participants on a scale from never to always, how often they vote in elections |
The project seeks to use the features above to estimate the likelihood that someone is a frequent voters.
Because this survey is forcing participants to answer qualitative information quantitatively, it has limitations. Many questions are asking participants abstract concepts and forcing them to narrow their opinions down to 2 or 4 options. This may limit their ability to accurately answer a question.
In addition, this dataset only contains approximately 6,000 responses. Compared to the US population of over 300 million, this dataset may not be an accurate representation of the US population. Despite this, it may still be helpful in identifying general attitudes.
We will use machine learning classifiers to predict based on certain opinions and values, whether or not someone will vote. We will also determing which factors play the most signigicant role in predicting whether or not someone will vote.