AP Research: Finding the Correlational Coefficient Between Support for Political Violence and Disbelief in the Electoral Process
Advisor: Mr. Hirsch
AP Research
This project will look for the correlation in support for political violence and disbelief in elections in Columbia and Georgetown students. Students are the subjects because they are the future. These are going to be future voters and leaders and political activists so by looking at them now will provide a rough estimate of what later general population studies will see can be determined. By using two schools, one in Washington D.C. and one in New York City the effects on witnessing political violence, the Capital Riot, on the student’s opinions can also be determined. When it comes to both support for political violence and belief in voter fraud, most of the current academic research is rooted in partisanship.
Project Blog
Week 0: AP Research Week 0: Introduction
Hi everyone. I’m Hudson, an AP Research student studying the relationship between how college students feel about elections and how that correlates to how supportive they are of political violence. I am specifically looking at NYU and Georgetown University. This is the area I chose to research because I love politics and the way it can help me understand this country in the same way physics does this universe. In these increasingly conspiratorial times, I wanted to study how these people who had no faith in elections act. I love hearing about groups like Qanon, initially because I thought they were funny and now because I view them as perhaps the single greatest threat to this country. I am currently conducting research and hope to finish within a few weeks. I am looking to find some interesting results and what they imply for this country. You’ll hear from me soon. Thanks so much, Hudson
Week 1: AP Research Week 1: The Capitol Riot
Hi, it’s me again. My whole project started after the 2021 Capitol Riot, where it was made clear how people’s lack of faith in the democratic institution could lead them to extreme measures. The correlation between this violence and a lack of faith in our election is shown by its motto, “stop the steal.” In an interview with CBS’s Face the Nation, former Trump National Security Advisor Fiona Hill showed the legitimate concern held by experts in this field that this could only be the beginning, referring to the event as a “dress rehearsal” (Watson, 2021) for things to come. After the Winter Palace-esque storming of our Capitol building, support for these rioters didn’t quell. The Republican Party defended the action, calling it “legitimate political discourse,” all but confirming Hill’s fear that this isn’t the end.
In addition, few conservatives and not nearly the expected amount of liberals call for the arrest of these individuals. After a 23% drop off between March and September, only 27% of Republicans view the prosecution as very important versus 80% of Democrats (Pew, 2021). People no longer see what was done as “that bad,” which could spell disaster for the future of politics. If events like the riot become a part of America’s political discourse, a greater understanding of the causes and solutions to such events will be needed. Hudson
Week 2: AP Research Week 2: The Gap In The Research
Last week I spoke about why I generally chose this area, but I really narrowed it down by looking at what gaps were left by other researchers. This study fills these gaps in three ways. First, pre-existing research fails to link political violence and faith in elections. Sometimes political violence will be linked to partisanship (as noted in “Dark triad, partisanship and violent intentions in the United States” or “The Correlates of Discord: Identity, Issue Alignment, and Political Hostility in Polarized America”). A lack of faith in elections is linked to a variety of factors like bureaucratic failures (see “Election Administration and Perceptions of Fair Elections”) or just conspiratorial thinking (like in “The Effect of Conspiratorial Thinking and Motivated Reasoning on Belief in Election Fraud”). The only study to really mention them both is the Survey Center on American Life’s look at a variety of post election topics. However, this survey is so broad encompassing every topic like patriotism and Donald Trump, and barely touches on political violence and election fraud. Even here, though, it treats the two topics as separate, and doesn’t mention any correlation. Many other studies look at how one or the other is worsening. The most extreme probably being the 2013 Fairleigh Dickinson University’s PublicMind Poll on Conspiratorial thinking in the US which found that almost 63% of registered voters believe in at least one electoral conspiracy, from Bush having previous knowledge of 9/11 to Obama concealing documents about his background. This doesn’t include the people who have lost so much faith in the system that they don’t even bother to vote. Nonetheless, the lack of investigation into a correlation has yet to be filled.
Furthermore, all of these surveys are far too broad in demographic to see how location affects it, like proximity to The Capitol. In this study, both NYU and Georgetown University students are being looked at. Both schools are similar enough so far as to see if Georgetown’s witnessing of the Capitol Riot may affect their support of political violence. If witnessing political violence weakens support for it, then overtime, in a world where it becomes more common, it will also garner less support. This correlation could mean that the problem solves itself.
Lastly, as this study focuses on the future, it focuses on a younger demographic. By choosing college students, not only does that help create consistency for the checking of proximity, it also helps look at how future voters and the political force of the future will act and feel. These students are future activists, voters, and politicians. By looking at them now, modeling their behavior in 15 or 20 years will be much easier. In order to fill the gap in these three ways, this study asks the question: What is the correlation in support for political violence and belief in unfair elections among Georgetown University and NYU students?
Week 3: AP Research Week 3: Partisanship
America is becoming more partisan than ever. According to the American National Election Studies (ANES), an academically-run government funded national surveys of voters in the United States, conducted before and after every presidential election, the amount of people who strongly identify as partisan is higher than ever before. Even those who claim independence from partisan politics are becoming a myth as the floating voter disappears. Corwin D. Smidt found for the American Journal of Political Science that the independents of 2016 acted like the most partisan of 1960, meaning that the flip flop voters of the past are all but gone. This is the first step in the problems of partisanship being accentuated.
At the same time, “both major parties are less popular than at any time in recent history.” In a 2018 copy of Advances in Political Psychology, Alan Abramowitz and Steven Webber argue the cause of this problem is “negative partisanship—the phenomenon whereby Americans largely align against one party instead of affiliating with the other”. This has led to a steep increase in disdain for the other party rather than an actual increase in like for one’s own party. This has accentuated the feeling of disdain out of party which could increase the apathy and violent tendencies the parties hold towards each other. If the parties fail to view each other as legitimate, then it only makes sense that they would view violence against each other as okay, as there would be no reason why they should care for those on the other side.
This is made clear when it comes to voting. Charles Stewart III, Stephen Ansolabehere & Nathaniel Persily found for the Stanford Law Review that opinions on the need for voting security were now based almost entirely on party lines. Partisanship is now a deciding factor in how elections are run. At this point in time partisan identities play such a massive part in any political discussion even just in the distinct ways the different parties act. In order to properly analyze the correlation the partisan distinction needs to be made.
Week 4: A Declining Faith In Elections
Conspiracy theories are becoming ever more prominent in American politics. A study that appeared in Fairleigh Dickinson University’s PublicMind poll, a think tank member of the American Association for Public Opinion Research (AAPOR) Transparency Initiative, found that a majority of voters believed in at least one election conspiracy, often including one side or the other committing voter fraud. This proclivity has only worsened since then. The Survey Center for American Life, a part of the American Enterprise Institute for Public Policy Research, a right-leaning high profile non-profit, found a large partisan divide in voter fraud belief with almost 100% of Democrats believing the election was legitimate, and only a third of Republicans agreeing. Compared to the PublicMind polls assessment of the 2004 election in Ohio, the number of Democrats who believed in fraud was cut in half, and the number of Republicans who believed so doubled. This may be in part due to the winner effect.
It was found that supporters of the winning candidate had much higher faith in elections than the losing party, so much so that it mitigated even pre-election fear-mongering. In the opposite vein, the pre-election fear-mongering that was mitigated in 2016 was probably strengthened in 2020 due to the opposite party’s victory. This may lead to overall less faith by some conservatives in the results of this study.
There is also a general lack of faith in the running of the system. Current perceptions that view the democratic process as poorly run decrease faith in it. With these compelling reasons for decreasing faith, it is clear to see how support for elections is only getting lower.
Week 5: Increasing Support For Political Violence
This week I will be talking about political violence. As partisanship increases, so does support for political violence. Partisanship and issue alignment, which in the US is almost perfectly aligned, were correlated with political hostility. This is backed up by research that found that stronger partisan identities drive stronger intentions to engage in political violence, though action is only found within those with certain acceptable personality traits. This last bit is countered by the large congregation of Americans who stormed The Capitol and brought into question to what degree these traits are needed.
The Survey Center for American Life’s same survey also measured this support for political violence. When asked if they agree with the statement: “The traditional American way of life is disappearing so fast that we may have to use force to save it,” 36% of respondents said yes. This shows remarkably high support for political violence. This number was highest among Republicans and lowest among Democrats which suggests more violent Republicans, a bias in the question, and/or the winner effect. There is definitely bias here due to the conservation of American values being more along the Republican party ideals but likely not as large a difference as was found. Support for political violence is high and rising. If electoral mistrust becomes a known and studied cause, dealing with it will be easier. See you next week!
Week 6: Methodology
I am using a survey to answer my research question. The two relevant strengths of an online questionnaire to gather data are large distribution and honesty. Honesty in particular is a benefit for me as people often want to give a socially acceptable answer, which would massively skew my data. By providing them anonymity through the internet I am able to gain more honest answers.
Surveys also have many disadvantages. Two in particular must be overcome in order for me to get accurate results, misinterpretation and low return rate. In terms of misinterpretation, I have dealt with that by making all relevant questions multiple choice, as well as putting my email at the top. By putting my email at the top, I can be contacted with any questions or such to clear up any confusion. In terms of my questions, I have made all but two of them multiple choice. By doing so I don’t allow for such confusion as they will see the possible answer which will help them derive what the question means. The three exceptions are “Who’s your favorite teacher?”, “If you align with a third party, please identify it here:”, and “What is your major/graduate program?” Where the first more serves the purpose of allowing my participants to get more comfortable with the survey, and the second is gathering specific demographic information, the third is incredibly specific and leaves little room for interpretation. While I may be able to mitigate misinterpretation through my questionnaire, I will have to deal with low return rates via action; I hope to do this through the “snowball method” of data collection. Because of this method I could get a return rate greater than 100% but, I predict that each send out will get me no more than 20 data points with diminishing returns with each email sent out. All in all this method matches what I need from a methodology so perfectly, it is nigh impossible to use a different method to better achieve my data.
This method entails me reaching out to a few individuals and then asking them to reach out to their friends and peers and their friends and peers etc. This both solves for low return rate, as people are far more likely to take online surveys if asked by their friends, and contributes to the large output that digital distribution is known for. This is reliant on my knowing someone to use as a base for the snowball, and I know a few people at the observed schools. Worst case scenario, if my connections fail to get back to me I can connect with other students whether it be through social media or email correspondence.
My method of choice does come with an additional flaw though. If I’m asking people through their peers, then I will likely get demographically similar participants. While this isn’t as large a problem as it can be as I measure correlation in a divisive demographic category, if I get to people who are of a certain opinion, then they will accurately reflect the data of those people. In addition, by asking some non-invasive demographic questions like major and political party, I will be able to weight the data based on school or general demographics, meaning I will value their correlation more, treating them as if they were more people since that opinion was under-represented. Through my survey, I am able to gather large quantities of data, and not worry about the many flaws that plague most surveys.
Week 7: My Results
I got more data back than I expected from a more diverse group than I expected. In total, 87 data points were gathered, from a good mix of demographics. There were 49 NYU students and 38 Georgetown students from a mix of political parties, including 11 third party voters and 22 independents. In addition, a majority were voters with 44 students.
In order to gauge the relationship found between the different indicator questions for faith in elections, I charted all of the relationships and calculated both slope and correlation coefficient. Among all of this though, only 5 relationships didn’t have a weak correlation as determined by less than a .3 Pearson product-moment correlation coefficient. The 5 that stand out are the Voter Fraud 2016 question for all students, Conservative students, NYU students, and Georgetown students as well as the Voter Fraud 2020 question for Conservative students only. The all and NYU students for the Voter Fraud 2016 question had a moderate Pearson product-moment correlation coefficient with a coefficient between .3 and .5. The other 3 non-weak correlations had strong coefficients with scores between .5 and .7. All of the non-weak coefficients were negative except Voter Fraud 2020 among Conservative students. The largest Pearson product-moment correlation coefficient was -0.67 for conservative students. My results weren’t as strong as I expected and there was more of a party divide than I thought, but overall I am happy with the results. See you next week!
Week 8: Implications
The numbers and correlation I found shows a lot about the political future of this country. The results of this study don’t bode well for a future free of political violence, though better than the initial hypothesis. The positives shown by the results were that there was a lack of correlation for liberals in all cases and voter suppression in all cases. Cracking down on making sure liberals view elections are fair and that there is transparency to show that there isn’t voter suppression isn’t necessary for legislators, election boards, or law enforcement. This is especially good as voter suppression has been getting more mention in the news due to gerrymandering news in states like New York and Florida.
The positives are limited though as what came from the school difference and conservatives imply more work needs to be done. While it may seem innocuous at first that there is no school difference, it shows that proximity to acts of political violence and dealing with its repercussions do not lower one’s score for political violence. This means that if political violence is on the rise, people will not react against it. Direct action must be taken to combat support for political violence as seeing how bad it can get and what it can mean for one’s community is not enough.
Most concerning though is how conservatives feel. It shows that belief in conspiracy relating to elections is a large factor in how they feel about political violence. It shows that results of studies like the PublicMind poll pose a large danger to this country as conspiracies are no longer in the realm of just believing in stuffed ballot boxes but to ideas of the deep state rigging elections in such a way that violence is the only solution. While liberals may not have to be targeted, conservatives must be shown transparency in the electoral process in order to make sure they know that elections are not rigged. If election boards, legislators, and law enforcement can keep conservative faith in elections high either through transparency or simply teaching about the electoral systems in schools, political violence can be limited. These implications aren’t perfect though. There were some serious limitations that I will address next week. See you then, Hudson
Week 9: Limitations
No study is perfect. Some of the unique design choices may have benefitted my study in some ways but harmed it in others. The winner effect playing such a large role in the results may be a downside. The 2020 election may have made conservatives more accepting towards violence and liberals less accepting. In the same vein, right wing media’s support for the Capitol Riot vs. left wing media’s condemnation may have led to diverging opinions. While this limitation may affect total results, it doesn’t affect the lack of correlation among liberals. Overall party rhetoric like that in the media may play an important role in establishing correlation, especially if that news source argues that elections are rigged and advocates for violence.
This survey’s rhetoric also impacted results. By asking the John Locke question, the violence was looked at in a more patriotic way and may have limited anti-patriotic political violence being supported. Anti-American sentiment is something that is often thrown around in politics as an accusation but the reality of it is more uncertain. Nonetheless this study failed to check for it due to how the questionnaire couldn’t argue that political violence is both patriotic and anti-patriotic without damaging the results.
In addition, the two schools being similar despite slight demographic differences may have impacted the results.
Campus and local culture may have impacted how people view these topics especially from two East Coast big city schools. While this may not interact with the specific research question, it may weaken this study’s ability to be more representative of all college students. In spite of this, it is unclear how campus culture may impact the correlation and not just the political makeup of the student body.
Lastly, this study falls victim to the same problem as all other studies looking at support for political violence as it doesn’t actually measure likeliness to commit political violence. This is an overall flaw in the field due to a general difficulty in measurement. Talk to you next week,
Week 10: Future Research
Hi everyone. This is last conclusion I reached in my research. Over the next few weeks I will instead talk about the presentation I did as well as reflect on this project as a whole. That being said all of that can begin by looking at what’s next. By doing the same experiment after a different party wins, it can be seen both how this correlation may shift over time, as even now it is likely becoming more extreme. This study would tackle two problems at once measuring the change in correlation allowing the field to get a timeline of extremity as well as mitigate for the winner effect.
The limitation of patriotic context for political violence in this survey can be solved by alternate surveys using similar wording. By removing the John Locke question or even replacing it with a question that argues that political violence may be anti-American, it may get different results and will help show how different ideologies supporting political violence may interact with elections.
To mitigate the similarity of the schools, this study could be expanded in order to gauge for more diverse student bodies. To establish how much one’s environment and non-party demographics play a role in establishing such a correlation or even support can be important for establishing those most at risk of committing political violence. In the same vein non-college students should be looked at because age is a key demographic missed by both this study and pre-existing literature. Non-students should also be looked at because people who do not go to college are an important voter base, especially for conservatives, and it is important to see how they may feel differently about such correlations.
One factor that likely established this correlation is both social and traditional media. It should be investigated how both quantity and source of consumption may establish how much of a correlation there is. Since the media played such an important role in the planning of the January 6th riot, as well as the partisan fight that followed, it is key to find how it is dominating people’s lives.
Lastly, it is important to accurately gather how support for political violence may lead to acts of violence. Works like Westwood 2021 are just the beginning. What is likely needed is a study that tracks participants to see if they ever get arrested for acts of political violence based on demographics and rated support for political violence. This would fill in the field creating more concrete results for this study as well as others like the Survey Center for American Life. By creating hard results like that it would better enable legislative and legal action in the prevention of election related political violence. Thank you so much, Hudson.
Week 11: The Presentation
Hi everyone. In my initial AP presentation I had 15 minutes to convey the whole idea of what I was researching. This week, I am going to be talking a little about my design choices and why I chose to emphasize certain ideas. The most important and drastic difference between my paper and presentation was the background research. I complete removed the introduction and boiled down the whole literature to one slide. By cutting all this, I was able to better focus on the real world impacts of my results. I also used quite a few slides on establishing what my paper was. The audience for my presentation is less academic than the audience for my presentation. This lead to me having to use a few of my slides to explain things like my hypothesis the gap in my research and why that is important.
The rest of my presentation was split between my data and my discussion. My methodology, results, and conclusion slides made up my data, with the methodology acting as context for the other two. They laid out the number sides of what I found like all of my correlation and the answers the questions I put in my survey. My implications, limitations, and future sections made up my discussion. Here I turned my data into an understanding of our world. I explain my correlations and future research can expanded upon it.
Hopefully you can watch my presentation during the senior project presentations. Thank you so much for making it all the way here.