Minority Group Size Research Articles

Unbiasing fairness evaluation of radiology AI model

Fairness of artificial intelligence and machine learning models, often caused by imbalanced datasets, has long been a concern. While many efforts aim to minimize model bias, this study suggests that traditional fairness evaluation methods may be biased, highlighting the need for a proper evaluation scheme with multiple evaluation metrics due to varying results under different criteria. Moreover, the limited data size of minority groups introduces significant data uncertainty, which can undermine the judgement of fairness. This paper introduces an innovative evaluation approach that estimates data uncertainty in minority groups through bootstrapping from majority groups for a more objective statistical assessment. Extensive experiments reveal that traditional evaluation methods might have drawn inaccurate conclusions about model fairness. The proposed method delivers an unbiased fairness assessment by adeptly addressing the inherent complications of model evaluation on imbalanced datasets. The results show that such comprehensive evaluation can provide more confidence when adopting those models.

Kinship, heritage, and ethnic choice: ethnolinguistic registration across four generations in contemporary Finland

Abstract We studied how individuals’ ethnolinguistic affiliation relates to the ethnolinguistic structure of kinship in contemporary Finland, a society in which Finnish-speaking and Swedish-speaking ethnolinguistic groups have coexisted for centuries and mixed marital unions are common. Using multigenerational data from the population register, we determined how the ethnolinguistic registration of children born in 1990–2015 relates to three generations of ancestors. We created a family tree that links children to their parents, four grandparents and eight great grandparents. Our intention was to both map the ethnolinguistic background of young people and predict a child’s affiliation based on their ancestry. The data revealed that ethnolinguistic affiliation is a more fluid and complex feature than expected when assessed only through child and parental characteristics. We found substantial diversity in ethnolinguistic background within the Swedish-speaking minority group, while most individuals in the Finnish-speaking majority group had a uniform background. We identified three types of bias in the ethnolinguistic affiliation of mixed-origin children: a matrilineal bias, a kinship majority bias and a Swedish ethnic minority bias. The analyses advanced our understanding of how the size of minority groups can shrink even when most couples in mixed unions favour minority group affiliation for their children.

Therapeutic Use of Acetaminophen and Light to Moderate Alcohol: Are There Early Disparate Risks for Kidney Disease?

Introduction: Acetaminophen (APAP) is the most common medication taken in the United States. Using the 2003–2004 U.S. National Health and Nutrition Examination Survey (NHANES), the authors previously explored and reported the relationship of concomitant consumption of light to moderate alcohol (LMA) and therapeutic doses of APAP to early risk of renal dysfunction among adults in the United States. Statistically significant increased odds of renal dysfunction were noted among respondents who reported use of therapeutic doses of APAP and LMA by adjusting for hypertension, diabetes, and obesity. In this study the authors explored further on of potential disparities in the above relationship in the population. The relationship was verified with the 2009 Chronic Kidney Disease–Epidemiology Collaboration creatine-based estimated Glomerular Filtration Rate (GFR).Methods: Subgroup logistic regression analyses to assess disparities based on gender, race, age, education, and income were performed for renal dysfunction measured in terms of serum creatinine (SCr) directly as well as self-report using NHANES 2003–2004 data.Results: Early stage renal dysfunction, as determined by self-reports, and SCr and GFR values may occur among those who concomitantly ingested therapeutic doses of APAP and described alcohol use when compared to those who do not. Risks were more profound among females, particularly in minority racial groups, below legal drinking age of 21, and with household income below $25K.Conclusion: Potential risks for renal dysfunction are apparent in a disparate manner resulting in possible health inequity. Further research could increase the sample size of minority groups and specifically assess for effect modifiers that NHANES does not include for assessment.

Who Controls Criminal Law? Racial Threat and the Adoption of State Sentencing Law, 1975 to 2012

Threat theory argues that states toughen criminal laws to repress the competitive power of large minority groups. Yet, research on threat suffers from a poor understanding of why minority group size contributes to social control and a lack of evidence on whether criminal law is uniquely responsive to the political interests of majority racial groups at all. By compiling a unique state-level dataset on 230 sentencing policy changes during mass incarceration and using data from 257,362 responses to 79 national surveys to construct new state-level measures of racial differences in punitive policy support, I evaluate whether criminal sentencing law is uniquely responsive to white public policy interests. Pooled event history models and mediation analyses support three primary conclusions: (1) states adopted new sentencing policies as a nonlinear response to minority group size, (2) sentencing policies were adopted in response to white public, but not black public, support for punitive crime policy, and (3) minority group size and race-specific homicide victimization both indirectly affect sentencing policy by increasing white public punitive policy support. These findings support key theoretical propositions for the threat explanation of legal change and identify white public policy opinion as a mechanism linking minority group size to variation in criminal law.

Testing a Theoretical Model of Perceived Audience Legitimacy: The Neglected Linkage in the Dialogic Model of Police–community Relations

Objectives: Democratic policing involves an ongoing dialogue between officers and citizens about what it means to wield legitimate authority. Most of the criminological literature on police legitimacy has focused on citizens’ perceptions of this dialogue—that is, audience legitimacy. Consequently, we know little about how officers perceive their legitimacy in the eyes of the public and the antecedents of such perceptions. Pulling together separate strands of literature pertaining to citizen demeanor, hostile media perceptions, and danger perception theory, we propose and test a theoretical model of perceived audience legitimacy. Method: We conducted two separate studies: the first a survey of 546 officers working at a southern U.S. agency and the second a survey of a national probability sample of 665 executives and high-ranking officers. Results: Local violent crime rates, but not minority group size or growth, are associated with lower perceived audience legitimacy. Additionally, recent experiences with citizen disrespect and global perceptions of citizen animus are both inversely associated with perceived audience legitimacy. The perceived hostility of local, but not national, media coverage is also associated with lower perceived audience legitimacy. Conclusions: Our results suggest a need for additional research that explores whether the antecedents of audience legitimacy indirectly affect police behaviors, like the use of force.

Homophily and minority-group size explain perception biases in social networks.

People's perceptions about the size of minority groups in social networks can be biased, often showing systematic over- or underestimation. These social perception biases are often attributed to biased cognitive or motivational processes. Here we show that both over- and underestimation of the size of a minority group can emerge solely from structural properties of social networks. Using a generative network model, we show that these biases depend on the level of homophily, its asymmetric nature and on the size of the minority group. Our model predictions correspond well with empirical data from a cross-cultural survey and with numerical calculations from six real-world networks. We also identify circumstances under which individuals can reduce their biases by relying on perceptions of their neighbours. This work advances our understanding of the impact of network structure on social perception biases and offers a quantitative approach for addressing related issues in society.

Understanding the Relationship Between Relative Group Size and Hate Crime Rates: Linking Methods with Concepts

In recent studies, Hipp et al. caution that the selection of the appropriate denominator in computing the rate of intra- and inter-group interactions is consequential for key findings. The present study builds on this work and examines whether adjusting for the structural opportunities for any type of interaction affects the observed relationship between hate crime rates and minority group size. We go beyond prior research by computing distinctive measures of anti-Black hate crime rates across US counties circa 2000. Our findings offer support for the power-differential hypothesis, revealing a negative effect of minority population size on crime motivated by bias. These results also underline the importance of the procedure developed by Hipp et al., showing that that the selection of the “baseline” for the computation of hate crime rates is critical for understanding the relationship between minority group size and crime motivated by bias.

Right‐wing adherence and objective numeracy as predictors of minority group size perceptions and size threat reactions

AbstractRacial, ethnic, religious, and sexual minority populations are growing rapidly in the West. We investigate predictors and consequences of threat responses to perceived minority group size, and perceptions of minority group size itself. Study 1 (N = 274) finds that right‐wing adherence predicts greater size threat, whereas lower numerical ability predicts larger perceived group size (confirmed in Studies 2–3). Study 2 (Sample 1, N = 124; Sample 2, N = 263) finds that the relation between right‐wing adherence and prejudice toward minorities exhibits an indirect effect via group size threat (but not perceived size), explaining 35–66% of this relation. Study 3 (N = 310) finds a comparable indirect effect explaining 27–40% of this relation, even after statistically controlling for indirect effects of realistic, symbolic, and terroristic threat. These findings provide novel insights into psychological processes surrounding perceived minority group size, identifying size threat as especially crucial in understanding intergroup relations.

Social Capital, Racial Context, and Incarcerations in the American States

AbstractThis article examines the differential effects of social capital on policy equity in state outcomes. Specifically, it explores the relationship between social capital and incarceration rates in the American states paying particular attention to racial disparities in incarceration rates. Building on work by Hero, I present a theoretical explanation and empirical support for how social capital operates differently under different racial contexts. I argue that social capital enhances social empathy in homogeneous contexts and social controls in diverse contexts. Using state-level longitudinal data on the contiguous states, I find that social capital is positively associated with incarcerations, but only for African Americans. Furthermore, the effects of social capital appear to be conditional on racial context where this relationship is stronger as minority group size increases.

Perceptions of Racial Group Size in a Minority-majority Area

Americans overestimate the size of minority groups and underestimate the size of the majority. Research on perceived racial group size, however, has focused on areas where non-Hispanic whites are dominant and has neglected to examine the impact of group conflict attitudes other than perceived threats. This study examines perceptions of group size in Harris County, Texas, in 2007, which became a minority-majority area by 2006. It also examines whether perceived racial discrimination influences perceived group size. Analyses demonstrate that people underestimate the size of the largest group, which in this case is the local Hispanic population. The sizes of the black and Asian populations, by contrast, are overestimated while estimates of the white population are accurate. Race and perceived discrimination play important roles in shaping perceptions of group size. Perceptions of group size are also based on the objective size of these groups, but are unrelated to recent changes in group size.

The Muted Consequences of Correct Information About Immigration

Previous research shows that people commonly exaggerate the size of minority populations. Moreover, as theories of inter-group threat would predict, the larger people perceive minority groups to be, the less favorably they feel toward these groups. Here, we investigate whether correcting Americans' misperceptions of one such population -- immigrants -- affects attitudes toward this group. We confirm that non-Hispanic Americans over-estimate the percentage of the population that is foreign-born or that is in the U.S. without authorization. However, in four separate survey experiments, we find that providing accurate information does little to affect attitudes toward immigration. This is true even when people's misperceptions are explicitly corrected. These results call into question a potential cognitive mechanism that could underpin inter-group threat theory. Misperceptions of the size of minority groups may be a consequence, rather than cause, of attitudes toward those groups.

How population structure shapes neighborhood segregation.

This study provides a framework for understanding how population composition conditions the relationship between individuals' choices about group affiliation and aggregate patterns of social separation or integration. The substantive focus is the role of income inequality in racial residential segregation. The author identifies three population parameters--between-group inequality, within-group inequality, and relative group size--that determine how income inequality between race groups affects racial segregation. She uses data from the Panel Study of Income Dynamics to estimate models of individual-level residential mobility and incorporates these estimates into agent-based models. She then simulates segregation dynamics under alternative assumptions about (1) the relative size of minority groups and (2) the degree of correlation between race and income among individuals. The author finds that income inequality can have offsetting effects at the high and low ends of the income distribution. She demonstrates the empirical relevance of the simulation results using fixed-effects, metro-level regressions applied to 1980-2000 U.S. census data.

The Politics of Race and Voter ID Laws in the States

Does partisan and racial context have an effect on the likelihood that states will adopt stringent requirements for voting? Our duration analysis shows that Republican governments increase the likelihood that a new law requiring citizens to have a photo ID to vote will be passed. This effect is weakened by minority group size. We then examine whether the adoption of voter ID regulations affects turnout across racial groups. Our analysis, using state-level data and the Current Population Survey (CPS) November Supplement File (NSF) for 1980 to 2010, offers little evidence for the belief that minority turnout is uniquely affected by voter ID regulations.

Racial Residential Segregation and Social Control: A Panel Study of the Variation in Police Strength Across U.S Cities, 1980–2010

Despite a great deal of theoretical and empirical attention given to racial residential segregation and its influence on a number of social problems in the United States, few scholars have examined the role that this persistent form of racial inequality plays in shaping the magnitude of formal social control efforts. Our study examines this relationship by assessing the potential influence that the isolation of minorities may have on efforts to control crime in urban centers across the United States. Using a pooled time-series regression technique well suited for the analysis of aggregate, longitudinal data, we assess the potential influence of racial segregation on the size of municipal police departments in 170 U.S. cities between 1980 and 2010. After accounting for minority group size, economic threat, crime, and disorganization, we find that racial residential segregation has a significant non-linear effect on police force size. Cities with the most racially integrated populations have the smallest police presence but at very high levels of segregation, police strength levels off. This finding is consistent with expectations derived from the contact hypothesis. Under such conditions, majority group members appear to be less inclined to demand greater crime control measures such as increased police protection. Period interactions with residential segregation also suggest that this relationship has grown stronger in each decade since 1980. Overall, our study provides strong support for threat theories and the contact hypothesis but offers necessary refinements.

An Agent-Based Simulation of Destigmatization (DSIM): Introducing a Contact Theory and Self-Fulfilling Prophecy Approach

In this paper, we propose a novel approach to exploring the destimgatisation process, an agent-based model called the destigmatization model (DSIM). In the DSIM, we demonstrate that, even if individual interactions (intergroup contact) are based on rules of the self-fulfilling prophecy, it can lead to destigmatisation. In a second study, we empirically verify a prediction that there is a positive relationship between minority group size and perceived stigmatization. Finally, we confirm that DSIM successfully implements Allport's (1954) four moderators, in turn decreasing the level of perceived stigma of the minority group. Interestingly, however, some of Allport's moderators influence the speed of destigmatisation, rather than having a lasting impact on the process of prejudice reduction. The findings suggest that moderators of 'intergroup contact' can function in one of two ways, either by improving how much contact helps to reduce stigmatization or by improving how quickly destigmatization can occur.

Racial Threat and White Opposition to Bilingual Education in Texas

This study examines local contextual conditions that influence opposition to bilingual education among non-Hispanic Whites, net of individual-level characteristics. Data from the Texas Poll ( N = 615) are used in conjunction with U.S. Census data to test five competing hypotheses using binomial and multinomial logistic regression models. Our results support a “racial threat” hypothesis, suggesting that increasing opposition to bilingual education among Whites corresponds to changes in the Hispanic population. We find opposition to bilingual education among non-Hispanic Whites to be most pronounced in areas with substantial growth in an already sizeable Hispanic population, and least pronounced in areas of high growth rates and historically low proportions of Hispanics. Importantly, our results highlight the relevance of the interaction between minority group size and minority growth rates in generating majority opposition to bilingual education programs in the United States.

Latino Immigration and Representation in Local Politics

Theories of racial politics argue that areas with large minority population are characterized by greater levels of policy equity, yet empirical research often fails to find evidence for this prediction. In this article, we develop and test an alternative view of the relationship between minority group size and the representation of minority interests in government using the case of Latino representation in local politics. We argue that Latino group size results in both favorable and unfavorable policy outcomes for Latino communities. The nature of the Latino population, specifically the extent to which it is composed primarily of citizens or noncitizens, explains when each outcome will occur. Latinos achieve more favorable policy outcomes for their interests when the size of the Latino citizen population increases but less favorable outcomes when the number of Latino noncitizens increases. The negative effect of noncitizen group size is mediated by the presence of Latino citizens and representatives. Our hypotheses are tested using data from a national sample of several hundred local school districts and educational policy outcomes.

Minority Group Size, Unemployment, and the Extreme Right Vote: The Crucial Case of Latvia

ObjectivesI test the importance of demographic and economic contextual variables for the success of extreme right parties.MethodsI employ subnational data from the crucial case of Latvia to capture the effects of minority group size and economic change on the extreme right vote. Ethnic Latvians constitute a bare majority of the Latvian population and GDP dropped by 18 percent in 2009.ResultsI find that localities with high or rising unemployment were less likely to support extreme right parties. The dampening effect of unemployment on the extreme right was greater in localities with large minority populations.ConclusionsThese counterintuitive findings raise questions about the logic and generalizability of existing explanations of extreme right voting. The study of the extreme right is hampered not only by ill‐suited measures of group size and economic conditions, but also by overarching theories that predict group‐based outcomes.

Racial Diversity and Public Policy in the States

How does racial diversity affect public policy outcomes in the states? The policy backlash hypothesis suggests that the presence of blacks increases antagonistic attitudes toward minorities among whites and thus produces racially conservative policies. The electoral constraint hypothesis argues that the presence of blacks increases the size of population that supports policy liberalism and thus results in more liberal policies. The authors reconcile these competing explanations by arguing that the hypothesized negative impact of minority group size on policy liberalism varies across states in accordance with socioeconomic and institutional conditions that strengthen or weaken the impacts of electoral constraint or backlash.

Valuing Diversity in Political Organizations: Gender and Token Minorities in the U.S. House of Representatives

Political scientists are keenly interested in how diversity influences politics, yet we know little about how diverse groups of political actors interact. We advance a unified theory of colleague valuation to address this puzzle. The theory explains how minority group size affects how members of a political organization differentially value majority and minority group colleagues, predicting that the effect of preference divergence on individual‐level colleague valuation is greatest when the minority group is smallest. We test this prediction using member‐to‐member leadership political action committee (PAC) contributions in the U.S. House of Representatives. The results obtain strong, albeit not uniform, support for the theory, demonstrating that the gender gap in colleague valuations declines as preference divergence increases in all but one instance. In contrast to conventional wisdom, the theory and evidence indicate that women serving in the U.S. House of Representatives receive less support from men colleagues as their ranks increase.