Cara Nix
Data Scientist & Python Enthusiast
Data Scientist & Python Enthusiast
Hi, I'm Cara! I am a data scientist and quantitative researcher with a PhD in political science and an MS in computer science from The Ohio State University. I also hold a BS in Mathematics from the University of Minnesota. I have experience with machine learning, natural language processing, network science, computer vision, and data visualization. I’m passionate about solving problems which benefit others. Previously, I worked for the Biden-Harris-Walz campaign.
The expanded role of technology in American politics has changed how legislators connect and share information with other legislators, constituents, and the mass public. By comparing and contrasting the formal legislative cosponsorship network for the 115th and 116th sessions of Congress with the more informal Twitter network of interactions between legislators for the same time period I will be able to determine if individuals that are ”powerful” or ”influential” within social media are also central to the legislative process. I will also determine if influence on social media makes one a more effective legislator. Exploring these relationships between legislative Twitter interactions and the cosponsorship network will demonstrate how accurately Twitter is reflecting the power structure and effectiveness of Congresspeople. The results indicate there is a positive and significant relationship between how influential a legislator is within the legislative network and legislative success.
Instances of police brutality have been shown to have a catalyzing effect on the electorate. For example, following the murder of George Floyd in Minneapolis in 2020 voter turnout in the 2021 mayoral election increased over 53 percent from 2017. However, despite a massive change in turnout indicating the mobilizing effect of police killings, Mayor Jacob Frey still won reelection in 2021. This puzzle of accountability is what I will explore further and put to the test at a national-level. Using data from police killings which occurred in medium and large (greater than 50,000 people) cities from 2013-2021 and corresponding mayoral election returns I consider the impact of police brutality on turnout, democratic vote share and incumbency vote choice. I expect to find turnout and democratic vote share in the mayoral election will increase following instances of police brutality. I also expect, relative to cities where there were no instances of police brutality, incumbency percentage will decrease, suggesting mayors are being “punished” for police brutality happening on their watch. However, I expect this effect is not large enough to unseat the incumbent, demonstrating contradictory outcomes. Increased mobilization, yet incumbent reelection- indicating a lack of mayoral accountability. I leverage the as-if random timing and location of instances of police brutality to implement a differences in differences design spanning multiple time periods. I also run a supplementary mediation analysis using local awareness as a mediator. Results indicate marginal effects of police killing on turnout, democratic vote share and incumbency support. This research highlights the limitations of elections at holding elected officials accountable for police brutality.
Utilizing cutting-edge geospatial and redistricting technologies, I investigate the impacts of redistricting at the school district and attendance zone level. First, by analyzing the existing boundaries and then by generating alternative districting plans. I explore if current boundaries are giving a specific racial group an educational advantage over another. This educational gerrymandering has normative implications as well as empirical realities for student achievement disparities within a state. I focus on educational boundaries within Minnesota, specifically, 79 school districts and 331 attendance zones. Comparing these geographic realities to alternative computer-generated boundaries, I find that, due to histories of red-lining and segregative politics, the existing attendance zones are less racially representative of the underlying population than those computer-generated. A statewide analysis of these educational boundaries allows for data-rich analyses of racial disparities within Minnesota’s education system.
Legislators communicate with constituents for a multitude of reasons. For example, campaigning, fundraising, and informing them of bills they are working on, or issues they are passionate about. Prior to the late 1970s it was difficult for legislators to communicate with their constituents, let alone the mass public. With the introduction of C-SPAN televising House floor debates in 1979 as well as the introduction of social media platforms such as Twitter in 2006, this provided legislators with new ways to communicate with their constituents and new data sources for analysis. This project explores variation in sentiment and emotion across the 115th, 116th, and 117th Congresses which includes a period of Republican unified control (115th), split government (116th), and Democrat unified control (117th).
In eight states, a “nesting rule” requires that each state Senate district be exactly composed of two adjacent state House districts. In this article, we investigate the potential impacts of these nesting rules with a focus on Alaska, where Republicans have a 2/3 majority in the Senate while a Democratic-led coalition controls the House. Treating the current House plan as fixed and considering all possible pairings, we find that the choice of pairings alone can create a swing of 4–5 seats out of 20 against recent voting patterns, which is similar to the range observed when using a Markov chain procedure to generate plans without the nesting constraint. The analysis enables other insights into Alaska districting, including the partisan latitude available to districters with and without strong rules about nesting and contiguity. Supplementary materials for this article are available online.
Used data from The Violece Project to create an interactive visualization of instances of gun violence in the U.S. over time using streamlit.
We sought to use computer vision to create a beauty bot that can be used to change different aspects of an image to enhance its aesthetic appeal to the user. The beauty bot includes the following tools, red eye removal, acne removal, retroactive portrait mode, and lip staining. This idea originally appealed to us due to its simplicity and due to the fact that computer science is a male dominated field. Therefore, most examples of computer vision are video games, virtual reality, robotics, etc. However, there are broader applications for vision and ones that would possibly encourage young women to consider careers in STEM. In the following section we will go through all the features of our beauty bot and discuss the vision techniques used as well as the success of the tool. Completed with Olivia Ridge.