I am a current PhD candidate at the Annenberg School for Communication at the University of Pennsylvania. I work with Yphtach Lelkes and am a member of the Democracy and Information Group and the Digital Media, Networks, and Political Communication Group. My current research follows two broad trajectories. The first occurs at the intersection of computational social science, political science, and cultural studies and investigates the structure and effects of relationships between political identity and cultural practices. The other centers around the role of digital platforms in online marketplaces for news and other media. My dissertation assesses the relationship between partisanship and cultural taste, employing both experiments and observational network analyses.
While I did not have much of it, I spent my time outside of academics during my undergraduate years contributing to SUNY Geneseo's cross country and track and field teams. In my undergraduate years, my personal bests were: 25:14 for the 8k, 15:34 for the 5k, and 31:20 for the 10k. In 2013, I ran at my first NCAA D3 Cross Country Championships. In 2014 and 2015, I had the honor of captaining the Geneseo Cross Country team. In 2015, I ran in my second NCAA D3 Cross Country Championships, helping Geneseo earn a 3rd place finish, the highest finish in the program's history. Today, while I do not run as much, I now devote some of my free time during the fall cross country season to collecting race results and using speed ratings to predict the team results at the D3 championships.
Culture and Partisanship
Missing Polarization: Using Networks to Compare Polarization in Political and Cultural Preferences
The popular press has given substantial attention to the notion that Democrats and Republicans hold diverging cultural and lifestyle preferences that manifest in the TV shows they watch, the music they listen to, and the clothes they buy. The academic research in this area is split, though, with some suggesting that such divisions exist and others arguing that they ultimately fail to materialize in real-world behavior. In this study, I use network methods to evaluate whether such partisan cultural polarization exists at the individual-level. I do so by constructing networks of shared cultural preferences and networks of shared political beliefs based on closed-ended survey responses. For each network, I calculate the assortativity (correlation) between linked respondents' partisan identity, ideology, age, gender, race, and education level. I show that the assortativity for the political identity measures is low across the cultural-preference networks compared to the political-belief networks. These results suggest that cultural preferences are not associated with partisan or ideological identities.
Partisan Overlap in Cultural Preferences: A Mixed-Methods Approach
Partisans are divided in their non-political preferences, leading to prejudiced interpersonal interactions. In particular, significant attention has been given to how partisans are divided by their entertainment-media preferences, which may indicate deep divisions in taste associated with partisanship. However, this divide may only be minimal and associated with the abandonment of preferences in the face of the group-identity misrecognition and competing influences. To test both possibilities, I asked partisans to recommend films every American should see in their lifetimes and to justify their recommendations. These recommendations were then the basis for constructing an audience network, for which I assessed fragmentation. Similarly, I coded respondents’ justifications and then modeled the occurrence of themes on respondent features. Results provide little support for the existence of deep cultural divides. Democrats and Republicans recommended many of the same films and used the same justifications for doing so, suggesting that arguments over cultural divergence may be overstated.
Partisanship or Culture? The Effects of Information Variety and Volume on Trust
Co-Authored with Do Eon Lee
There is little doubt over the existence of affective polarization, but findings on the related causal effects of party cues on non-political behavior may be affected by design decisions related to the volume and types of information used in experiments, as well as the trust framework employed. Here, we consider how these effects vary across low and high information environments for a less impactful trust context replicating initial trust conditions. We find that in low trust conditions, the effects of party cues are stable between party groups and are strong relative to the effects of other pieces of information about race, gender, religion, religiosity, policy preferences, and cultural preferences. However, such effects are reduced and become highly moderated by party affiliation within the high information environment. These findings suggest that differences between the implications of earlier research on the behavioral consequences of affective polarization and daily life may be explained by assumptions made in the design of earlier studies.
National news outlets are favored over local news outlets in news aggregator results
Local news plays an essential role in ensuring the healthy functioning of democracy. However, local news outlets have struggled to stay open in the more competitive market of digital media. The factors contributing to this struggle are not entirely clear. Demand-side preferences certainly play a role, but supply-side decisions may also divert readership in ways that are harmful to local news outlets. To gain a better understanding of how one major kind of gatekeeper -- an online news aggregator -- may be affecting the ability of media consumers to access local news outlets, we conduct an audit of Google News. While there is very little local news on the default portal, we find evidence that the amount of local news returned by Google News depends heavily on the actual query used and not geographic or market-specific features.
Local news availability does not increase pro-social pandemic response
The response to the COVID-19 pandemic in the United States has been notably partisan. However, recent evidence suggests that people have also been directing more attention to local newspapers during this period. Given that local newspapers promote pro-social civic behavior, such as turning out to vote, it is possible that this increase in attention is helping communities to adopt necessary social distancing behavior. To test this possibility, I combine data from Google on mobility in thousands of American counties with counts of the number of newspapers available in each county, as well as county-level pandemic and demographic fea- tures, to model changes in staying at home and traveling for retail and recreation purposes. I find that even though behavior change is corre- lated with local newspaper availability, the association disappears when controlling for additional pandemic and demographic features. The lack of an effect persists even when applying covariate balancing propensity score weighting. The lack of a causal effect of local news availability on social distancing uptake suggests that local news is limited in its ability to undo the politicization of national issues.
Causal Effect of Playing Time in the NBA
Recent shifts in professional basketball have led teams to place more urgency in drafting as well as possible. Draft picks must play out their initial years under team-friendly contracts that provide teams with increased salary cap flexibility. Yet, while this urgency has led to widespread discussion and research of how to improve teams' draft decisions, little attention has been given to identifying what teams can do to maximize the performance and potential of their draft picks once they are added to their roster. However, theories of learning and ecological psychology suggest that giving young players as much playing time as possible should lead to concrete improvements in their development and future performance. In this study, I test this causal theory by evaluating the relationship between the minutes a player receives in their first two seasons in the NBA and their fourth-year performance using a novel method of propensity score weighting that enables weighting for continuous treatment variables. I find that players who receive more minutes in their first two seasons have better fourth seasons and make larger jumps from their first two seasons to their fourth season, controlling for a broad set of potential confounders. These results have important implications for teams as they develop organizational strategies for the short- and medium-term.
- Data Visualization Portfolio
- PDF Copy of Portfolio:
- Who is the Worst NBA All-Star in Recent History/a>
- 2019 NCAA D3 Men's Cross Country Predictions
- 2019 NCAA D3 Women's Cross Country Predictions
- 2018 NCAA D3 Cross Country Predictions