Candidate viability scores developed by USC Professor Christian Grose—originally constructed to determine eligibility for a March 2026 gubernatorial debate co-hosted by USC, ABC, and Univision—demonstrate significant predictive power for California primary election outcomes. Though the debate was cancelled amid political controversy over the selection formula, we set aside this dispute to examine a broader empirical question: do objective, pre-election viability metrics, derived from candidates' polling and fundraising strength, predict primary electoral performance? Drawing on all California gubernatorial primaries over the past 20 years and all U.S. Senate primaries since 2010, we find that candidates with the highest viability scores consistently advanced to the general election. OLS regressions reveal that pre-election viability scores are robustly associated with primary vote shares, even after accounting for additional covariates. The 2016 U.S. Senate primary exemplifies this relationship, as the two candidates with the highest viability scores (Kamala Harris and Loretta Sanchez) advanced with vote shares closely aligned with their predicted standings. Applying this framework to the 2026 cycle, we identify a structural risk of Democratic vote fragmentation stemming from score clustering among a large candidate field, and find a non-trivial probability that two Republican candidates could advance through California's top-two primary system despite the state's Democratic-leaning electorate.

Latino-owned businesses (LOBs) are emerging as a transformative force in the U.S. economy. Representing nearly 10% of employer firms, these 553,000 businesses generated $832 billion in revenue in 2023, with growth in firm creation, employment, and revenue consistently outpacing White-owned businesses. Between 2017 and 2023, LOBs added 180,000 net new firms and nearly one million jobs. A growing share operate in technology sectors, expand internationally earlier than peers, and pursue acquisitions and new markets at higher rates. Despite this momentum, structural barriers persist: LOBs face chronic funding gaps that widen with loan size, receive less actionable feedback after denials, and have seen their venture capital participation collapse at the early stage, pre-seed/seed share dropped 79% since 2020. Addressing these capital access disparities through clearer lending guidance, targeted investment, and supportive policy will be essential to sustaining and scaling Latino entrepreneurship's considerable economic contributions.

Gerrymandering remains a central challenge to democratic representation in the United States, shaping both partisan advantage and the electoral prospects of historically underrepresented communities. This piece traces the evolution of redistricting law and methodology across two doctrinal strands: constitutional limits on partisan gerrymandering and statutory protections against racial vote dilution under the Voting Rights Act. It argues that contemporary redistricting disputes increasingly turn on the interaction between legal standards and empirical tools used to evaluate fairness. After reviewing early geometric measures of compactness and outcome-based partisan metrics, the article highlights the rise of computational ensemble simulations as the dominant framework for detecting partisan bias. It then examines race-focused methods, including ecological inference and emerging techniques for identifying racial predominance, alongside developing efforts to operationalize “communities of interest.” Taken together, these approaches reflect a broader methodological shift from intuition to data-driven inference, raising new questions about transparency, assumptions, and administrability in redistricting adjudication.

In the US Congress, a key signal of a legislator’s effectiveness is their sponsorship of bills, but minority representatives, particularly women of color, often face institutional and informal barriers to sponsoring legislation. In new research, I explore how minority representatives work around these constraints by cosponsoring bills. I find that newly elected minority members are likely to have a 20 percent higher cosponsorship rate compared to their white predecessors, and that this rate is even greater for legislators who are women of color.

Ranked Choice Voting (RCV) is frequently promoted as an alternative to Plurality Voting, with the goal of enhancing democratic representation. As its gained traction across various jurisdictions—ranging from states like Alaska and Maine to cities like New York and San Francisco—the impact of RCV on minority communities has become a critical area of research and debate. While the adoption and awareness of RCV continue to expand, scholars have increasingly raised concerns about its effectiveness in fulfilling its promises of improved representation and equity. Some of the literature has shown that RCV often falls short in addressing these goals, particularly for underrepresented minority groups. As such, there is a growing need to critically assess RCV's limitations and explore alternative voting methods that may more effectively promote fair representation. One such alternative is Approval Voting, which presents a promising approach to overcoming the shortcomings of RCV. Approval Voting could offer a more equitable solution, especially in terms of its impact on minority voters.

The America (Mis)Represented: 2022 Election Report investigates the prevalence and consequences of vote splitting in the United States’ 2022 election cycle. Vote splitting—where multiple candidates divide the vote in plurality systems—frequently results in non-majority winners, undermining representative outcomes. Analyzing 5,662 national primary elections, the report finds that 11.9% exhibited vote-splitting dynamics, with pronounced effects in states such as New Hampshire, Arizona, and California. Voter dissatisfaction has intensified, with public confidence in democracy dropping from 41% in 2021 to 31% by 2024. Logistic regression analyses reveal that open-seat and competitive races are significantly more prone to vote splitting. The report advocates for reforms such as approval voting, which mitigates vote distortion and better captures voter preferences. Evidence from simulations, including California’s 2024 Senate primary, suggests that such reforms could lead to more representative electoral outcomes.

Interdisciplinary research domains often study the same phenomena using incompatible terminologies and measurement frameworks, creating silos that impede cross-domain data integration. Existing approaches rely on manual crosswalks that do not scale or domain-specific ontologies that do not generalize. Large Language Models offer new capabilities for semi-automated ontology development, yet no existing framework connects LLM-assisted ontology construction to social-science concept mapping across disciplines and measurement instruments. We present a generalizable framework for LLM-assisted ontology development supporting cross-domain integration and scalable annotation. The framework uses a three-layer architecture: OWL 2 for formal concept hierarchies, SKOS for cross-disciplinary terminological mapping, and SHACL for data validation—making the approach both semantically rigorous and operationally deployable. We demonstrate the framework using procedural justice in policing, unifying operationalizations from criminology, computer science, organizational behavior, political science, and sociology under a shared ontological structure. This infrastructure enables organizations to adopt a common semantic backbone, compare operationalizations simultaneously, and conduct previously infeasible cross-domain analyses.

Encounters with the police link race, place, and class to political, social, and economic outcomes. Hundreds of studies document how individual and contextual factors predict exposure to police contact and its downstream consequences, what we term the contact-to-consequences chain. However, the nature of police-civilian interactions remains poorly understood because it has thus far been impractical to observe a random sample of interactions. We introduce data from human annotations of approximately 900 LAPD traffic stops drawn from body-worn camera footage, comprising over 13,000 hours of coded interactions. Using sixteen indicators of officer behavior, we construct a new index of the intensity of police-civilian encounters. We then link these data to administrative stop records, officer personnel data, and geospatial neighborhood context data. We show that “quality” contact with the state is distributed across people and places. We show how our measure can advance various fields, with and document this by providing insights on procedural justice, representative bureaucracy theory, and political control of police. Our results provide the first systematic measure of encounter severity from body-worn camera data and demonstrate how BWC footage can identify where and for whom the burdens of intense policing are most heavily imposed.

Mexican municipalities maintain official historical narratives that selectively remember the indigenous and colonial past, celebrating heritage in some cases, naming colonial harm in others, and in many cases doing neither. We argue these choices reflect two conceptually distinct orientations in official discourse: indigenous affirmation, the recognition and celebration of indigenous identity, culture, and heritage as living and valued; and colonial critique, the explicit acknowledgment and condemnation of colonial structures and their enduring legacies. Crucially, these orientations are not two sides of the same coin. A municipality may honor indigenous culture while remaining silent on the violence that shaped it. Drawing on the encyclopedic municipal records of over 2,400 Mexican municipalities compiled by the Instituto Nacional para el Federalismo y el Desarrollo Municipal (INAFED), we measure both dimensions at scale using a theoretically grounded classification battery and Item Response Theory. The resulting indices reveal geographic heterogeneity and surface a distinct pattern of heritage-without-reckoning that prior approaches to official historical memory have been unable to capture.

With the U.S. House of Representatives Republican majority hanging in the balance—President Trump has turned to redistricting as a possible means of maintaining GOP control in an upcoming 2026 midterm environment likely to favor Democrats. Political observers aptly characterize this latest partisan battle over congressional redistricting as an arms race. Indeed, with Texas Republicans passing a new map with the goal of netting five more seats, failed deterrence has come in the form of California Democrats responding in kind with a map voters approved, which is expected to net their party five additional seats. Further, due to legal provisions and litigation, two other states have new plans enacted for 2026, Ohio and Utah; and several other states have redrawn their maps in the hopes of partisan gains. In this paper, we examine all pre-2025 and post-2025 maps passed during this cycle of mid-decade redistricting. Using simulation methods placing various constraints on districts, we automate the drawing of 750,000s of maps in each of these states. We then compare and contrast the old and new map in each mid-decade redistricting state to evaluate for partisan bias, and determine which mid-decade redistricting maps are the most partisan outliers relative to a baseline simulated distribution of plans. We chronicle this latest round of “re-redistricting,” place it in historical context, and analyze these new maps to determine their effectiveness in meeting the line drawers' intentions.

Most audit studies of legislative responsiveness find that racial minorities receive less engagement from elected officials. Yet recent evidence from Mexico reveals the opposite: Mexican legislators exhibit pro-Indigenous responsiveness. This finding raises a critical theoretical question: does this responsiveness reflect genuine substantive engagement or merely symbolic performance? We use a large-scale audit experiment with all 626 members of the Mexican federal legislature to disentangle these two forms of responsiveness. Using block randomization stratified by chamber, party, and gender, each legislator receives nine constituent inquiries over nine consecutive weeks, varying both racial identity (Indigenous, Mestizo, European-White) and inquiry type (service-oriented, policy-oriented, and symbolic). This tripartite design allows us to test whether pro-Indigenous responsiveness extends to costly substantive domains or remains confined to low-cost inclusionary gestures. Responses are coded across four dimensions: reply presence, personalized greetings, warmth, and substantive engagement with the inquiry. We evaluate two competing hypotheses. The symbolic compliance hypothesis predicts that legislators overcompensate toward Indigenous constituents through performative inclusion but fail to follow through substantively. The substantive responsiveness hypothesis predicts that pro-Indigenous bias extends to meaningful engagement, even when politically costly. Results will shed light on how institutional context, party affiliation, electoral competitiveness, and legislator gender condition these dynamics. This study contributes to global theories of racialized representation by offering the first experimental test of whether descriptive responsiveness to marginalized groups reflects surface-level signaling or genuine political incorporation.

This study explores how top-two primary systems affect ideological diversity in U.S. state legislative parties, focusing on California and Washington. Using a synthetic difference-in-differences (SDID) approach, it compares ideological heterogeneity, measured with Shor-McCarty ideal points, across different primary systems. Findings show that top-two primaries significantly increase ideological diversity among lower chamber Democrats, consistent with greater same-party competition. Upper chamber Republicans also exhibit consistently positive and substantively meaningful effects, reaching statistical or marginal significance in most specifications. Results for lower chamber Republicans are directionally consistent but more sensitive to model specification and pre-treatment dynamics, warranting cautious interpretation. No significant effects are observed for upper chamber Democrats, underscoring how institutional features condition reform outcomes. The study highlights how designs like SDID enhance causal credibility in complex settings and shows that electoral reforms like top-two primaries can meaningfully reshape the ideological composition of legislatures—often in asymmetric and context-dependent ways.

How do race, ethnicity, and gender causally shape legislative behavior within institutions presumed neutral? This study advances a theory of strategic persistence to explain how marginalized legislators navigate compounded constraints—electoral vulnerability, institutional exclusion, and representational burdens. Drawing on theories of descriptive and substantive representation, intersectionality, and strategic behavior, I argue that race and gender jointly condition legislative influence. Using a difference-in-differences design with exact matching and negative binomial fixed effects models, I analyze House member activity from 1981 to 2022. Results show that minority legislators, especially women of color, sponsor fewer bills but engage in significantly higher cosponsorship patterns that persist over time, intensify under electoral pressure, and remain even with seniority. These findings reveal that while descriptive representation may open the door, it does not ensure equal institutional power. Instead, cosponsorship is used as a durable, strategic tool for navigating exclusion and building influence within unequal institutional structures.

Studies of political representation tend to focus on public preferences and legislative policy. This scholarship largely relies on operationalization of “policy” that elides how policies are implemented by street-level bureaucrats and perceived by individuals with varied experiences and expectations of the state. This approach obscures the possibility that individuals, even if they report being exposed to the same policy output through interactions with street-level bureaucrats, may nonetheless have divergent perceptions of implemented policy that, in turn, inform their political attitudes. To address this possibility, we combine survey and interview data about preferences and experiences with police traffic stops in Los Angeles, CA. Survey results show little variation across racial groups' reported policing preferences and prior treatment in traffic stops. Nonetheless, we observe considerable variation in views of police performance. Interview data indicate that this disjuncture between reported treatment and perceived performance is tied to minority respondent's feeling of fear and anxiety created by direct and vicarious past policing experiences. Our results underscore the utility of looking beyond the letter of formal policy to account for policy implementation in the study of public preferences and their practical representation.

Face-to-face interactions between police officers and the public affect both individual well-being and democratic legitimacy. Many government-public interactions are captured on video, including interactions between police officers and drivers captured on bodyworn cameras (BWCs). New advances in AI technology enable these interactions to be analyzed at scale, opening promising avenues for improving government transparency and accountability. However, for AI to serve democratic governance effectively, models must be designed to include the preferences and perspectives of the governed. This article proposes a community-informed, approach to developing multi-perspective AI tools for government accountability. We illustrate our approach by describing the research project through which the approach was inductively developed: an effort to build AI tools to analyze BWC footage of traffic stops conducted by the Los Angeles Police Department. We focus on the role of social scientists as members of multidisciplinary teams responsible for integrating the perspectives of diverse stakeholders into the development of AI tools in the domain of police and government accountability.

Police wield broad discretion. They are also agents of political principals who set policy that prescribes their behavior. We develop a model of police reform and argue that successful implementation requires that either (1) officers do not resist key policy elements or (2) principals are willing to sanction non-compliant officers. We test our theory using a 2022 Los Angeles Police Department reform regarding the practice of pretextual traffic stops. Drawing on body-worn camera footage and millions of stop records, we find that officers complied with the stop documentation reforms, which were not strongly opposed. In comparison, early reductions in pretextual stops dissipated within a year. Interviews with LAPD personnel support our model's predictions that noncompliance was driven by agent resistance and a lack of expected sanctions for policy violations.