Racial Indicators Research Articles

Implicit bias in large language models: Experimental proof and implications for education

We provide experimental evidence of implicit racial bias in a large language model (specifically ChatGPT 3.5) in the context of an educational task and discuss implications for the use of these tools in educational contexts. Specifically, we presented ChatGPT with identical student writing passages alongside various descriptions of student demographics, including race, socioeconomic status, and school type. Results indicate that when directly prompted to consider race, the model produced higher overall scores than responses to a control prompt, but scores given to student descriptors of Black and White were not significantly different. However, this result belied a subtler form of prejudice that was statistically significant when racial indicators were implied rather than explicitly stated. Additionally, our investigation uncovered subtle sequence effects that suggest the model is more likely to illustrate bias when variables change within a single chat. The evidence indicates that despite the implementation of guardrails by developers, biases are profoundly embedded in ChatGPT, reflective of both the training data and societal biases at large. While overt biases can be addressed to some extent, the more ingrained implicit biases present a greater challenge for the application of these technologies in education. It is critical to develop an understanding of the bias embedded in these models and how this bias presents itself in educational contexts before using LLMs to develop personalized learning tools.

Tooth loss over 13 years of follow-up: Can regular dental visits reduce racial and socioeconomic inequalities?

To assess if regular dental visits modify the effects of social and racial indicators on the incidence of tooth loss. This is a longitudinal analysis using data from the Pro-Saude Study. In 1999-2001, 3253 civil servants responded to self-administered questionnaires, and then in 2012-2013, with 19% attrition. The outcome was any increase in self-reported tooth loss, measured in four ordered categories (none, one or few, many, all or almost all). Main variables included income, education, race/ethnicity and an adapted version of Everyday Discrimination Scale. The dental visit was dichotomized into regular and problem-oriented attenders. Potentially confounding factors were age and sex; effect modification was estimated using the relative excess of risk due to interaction (RERI). An increase in the tooth loss category was reported by 23.1% of the individuals over 13 years of follow-up. Among problem-oriented attenders, 27.3% reported an increase against 20.4% in regular users (p<0.01). Interaction results are inconclusive. Even though not significant, either antagonism or synergism were observed: between lower income and problem-oriented (RERI=-0.22; 95%CI: -0.75: 0.31), being Black+Brown and problem-oriented (RERI=-0.25; 95%CI: -0.64: 0.14), discrimination and problem-oriented (RERI=-0.15; 95%CI: -0.55: 0.25), and between having less than university degree and being problem-oriented (RERI=0.21; 95%CI: -0.19: 0.62). Regular attenders from advantaged groups seem to benefit more from dental care than disadvantaged groups, increasing unfair inequalities. Inconsistencies in current findings warrant further investigations. Regular attenders from advantaged groups seem to benefit more from dental care than disadvantaged groups, increasing inequities, but the effect size of the Relative Excess of Risk due to Interaction was not large and was inconclusive.

Virtual Flânerie Teju Cole and the Algorithmic Logic of Racial Ascription

This article reads Teju Cole's award-winning novel Open City (2011) as an extended allegory for the operations of a 'technological unconscious' recently theorized by critics including Nigel Thrift, David Beer, Alexander Galloway, and Katherine Hayles. Cole is widely recognized for being an innovative social media activist; and yet Open City has almost uniformly been labeled an 'antiquarian' text, the highly lettered account of a Nigerian-German flâneur (Julius) who 'aimlessly wanders' New York in search of his racial identity. Yet an emphasis exclusively on Cole's 'antiquarian' style risks missing his novel's formal engagement with the technological present. Drawing on Lev Manovich's account of the flâneur as a figural precursor to the Internet user, I argue that the twenty-one short chapters that serialize Julius's 'aimless wandering' also chart his gradual slotting into an implicitly preferred racial identity category, as Julius is repeatedly hailed by African and African American persons and objects eager to gain his attention. Carefully yoking Julius's flânerie to the protocols of now-ubiquitous systems for user tracking and content personalization notably deployed by Internet browsers like Google, Open City subsequently shows how Julius's 'aimless' desire for serendipitous and diverse social encounters yields, instead, a subtly curated journey through New York cued by his perceived racial indicators. In this way, the novel presents the 'Open City' as a metonym for the 'open' Web in order to lay bare the ideology of 'openness' itself, giving the lie to pervasive cultural investments in the Internet as a 'postracial' global mode of production, and illustrating race's systemic drag within a current phase of tech-intensified capitalism whose putatively neutral proxies reiterate global class formations along racial lines.

Whose Fault Is It?: A Survey Experiment Examining Differences in Attribution and Support for White and Black Men

Race has an enormous impact on the mindset and collective culture of the United States. Previous research has established that race impacts housing, employment, healthcare, and educational opportunities, partially as a result of ideologies that favor internal attributions. The present research aimed to examine the relationship between attribution and race, and support for social policy and race, in addition to the impacts of level of education, field of study, social political orientation and race. To investigate these relationships, participants were randomly assigned to one of six conditions consisting of a vignette, racial indicators and a series of questions on Amazon Mechanical Turk. Results demonstrated that, in line with traditional American ideology, a majority of respondents made internal attributions for difficult life situations. Internal attributions, causative factors linked to personal characteristics or decisions, were more frequently assumed than external factors, the causative factors linked to societal structure. However, liberal respondents, white respondents and female respondents were more likely to make external attributions and to support social policy for black individuals than for white individuals. Social political orientation had the strongest relationship with both attribution by racial condition and support for social policy. The trend of external attributions and policy support for black individuals more so than for white individuals is noteworthy, and necessitates further research.

Disparities in Life Expectancy Across US Counties Linked to County Social Factors, 2009 Community Health Status Indicators (CHSI)

In response to a National Healthy People 2020 goal to create social environments to improve health, we examined US community health status by social factors to identify opportunities for health improvement. County-level relationships between life expectancy and known social determinants of health were assessed. An ecological study of all 3,141 US counties or equivalents using 2009 Community Health Status Indicators dataset was conducted. We examined health status and social indicators across counties and correlation of average life expectancy with educational, income, and racial indicators. Life expectancy varied across US counties (median, 77 years; range, 67–81) and was negatively correlated with percentages of residents living below the federal poverty level (r = −0.65; P < 0.05), of adults aged ≥25 years without a high school diploma (r = −0.64; P < 0.05), and of African–American residents (r = −0.56; P < 0.05). By contrast, life expectancy positively correlated with percentage of White-American residents (r = 0.51; P < 0.05). Correlations persisted across counties with varying poverty and educational attainment; strongest correlations found in counties with largest percentage of residents in poverty or adults with no high school diploma. Life expectancy across US counties is moderately to strongly linked with county poverty, educational attainment, and racial composition. Counties with the longest life expectancy need further examination to identify their effective policies, resources, and investments responsible for better health.

Buried Traces

By the time I came to make this film, my grandfather had already passed on. There was no one in my family to speak about their experiences growing up Métis. It was in talking to Lorraine Freeman, a Métis community leader in Winnipeg, that I came to really comprehend the extent of racism and challenges of being a ‘halfbreed’ in the first half of the 20th century. My conversations with Lorraine helped me to see that the years of silence around our culture and identity was linked to a much broader colonial legacy.&#x0D; &#x0D; When I was growing up if you said who you were, somebody was duking it out with you on the baseball diamond at recess. That’s just the way it was ‘cause we were nothing but dirty halfbreeds. My uncle told me one time he was speaking Michif to a friend of his and the nuns put his head down the outside toilet cause he was speaking this abomination of a language.(Lorraine Freeman)&#x0D; &#x0D; Lorraine was able to say what my family wasn’t. It is as if through this film, I was able to engage with Lorraine’s memory, to access the difficult experiences of my grandfather, to make sense of my family’s silence. Remembering the assault on the body, embodying this past, searching for a voice within this fraught identity.&#x0D; &#x0D; Almond shaped eyes, lack of the carabelli cusp on the maxillary first molar, a wide space between the big toe and the second toe ...Dispelling the myths of what "Indians" look like (Duffy 2010)&#x0D; &#x0D; For people who call themselves Métis, the ongoing battle of who belongs, the struggle for political inclusion, is played out in the contested body of the “hybrid halfbreed”. Who we are is framed between fragmented memories, memories denied, images, stories, rumors, and a legacy of colonialism’s crusade for assimilation if not extermination. In the early 19th century, the physiology of the inferior non Anglo-Saxon was his/her cultural identity, a racial classification wherein, as Shohat and Stam write, “a mania for classification, measurement, and ranking – expressed in such pseudo-sciences as phrenology and craniometry – left no domain untouched. Every detail was mastered in the name of abstract hierarchies” (1994: 91). This racism disguised as science was particularly vicious towards persons of mixed race, “dreaded by racists as a monster, an infertile hybrid” (Pieterse in Shoat and Stam 1994: 91). This view shaped the life experience of the Métis Nation in very concrete ways. Bonita Lawrence notes how colonial surveyors classified halfbreeds and Indians based on racial indicators (2004: 88). Those that “looked and acted Indian” were classified as Indians. Others assimilated into the white majority. Yet others of this “mongrel” race remained on the margins, surviving in isolated communities, many just outside the boundaries of the reserve. Some, pushed off their land base, scrip squandered, became the “road allowance people”, making a life and settling on the thin strips of undesignated land on either side of the roadways.&#x0D; I blur the boundaries between past and present – between Lorraine’s voice and my body, between historical footage and contemporary footage of me and my grandmother, between photos of Métis heroes and today’s prairie landscape. Contemporary lives and struggles are infused with the knowledge and experience of the past. I draw connections between current realities and the political repression and social exclusion that came before us. I layer images and suture fragments together. Our identities and memories are unstable and impermeable.&#x0D; I survey my own body: Do high cheekbones or almond shaped eyes make you an Indian? To what extent do lived experience, historical affiliation, family lore, blood quantum and physiology play into notions of cultural identity? Who decides? How does self-identification manifest itself in relation to policy decisions and struggles for rights and recognition?&#x0D; It is not my parents’ or grandparents’ experiences that I “remember”. They were silent about their trauma. Lorraine Freeman’s testimonial becomes my expression in a metaphorical sense, breaking down the division of past and present, infusing my own experience as a Métis person today with the knowledge and experience of this past. This “postmemory” invokes, as Lebow describes in reference to Chantal Ackerman’s work, “a broader, ancestral postmemory” (Lebow 2008 19) For me, Lorraine’s words reflect a collective experience that is life affirming and political. We are still here. We have survived.&#x0D; Buried Traces has been selected and featured in a number of recent festivals and galleries, including: Images Festival 2009, Gimme Some Truth, Festival Prescence Autochtones, GIV summer festival, Métis Media Festival, FIFA 2011, Traverses Festival Toulouse, Métis Health: Culture, Identity, History conference, NAHO; Galleries and Curated Collections: Alternator Gallery Kelowna, Urban Shaman Winnipeg, Topo-vidéographies: Territoires/Territorios Montreal. Copies of the work can be ordered via GIV http://www.givideo.org/ang/indexA/videosA/A-299.html.

House Price Differentials in U.S. Cities: Household and Neighborhood Racial Effects

Three cities are examined for evidence of housing market discrimination and prejudice using the American Housing Survey for 1978–1991, supplemented with information on neighborhood quality from the 1980 and 1990 Censuses. We show that incorrect inferences about the magnitude of racial price differentials are possible when neighborhood characteristics are excluded. When this information is included and the racial indicators are allowed to vary over time and submarkets, there is evidence of increasing prejudice in Denver and Philadelphia and decreasing prejudice in Chicago. There also appears to be a trend toward lower levels of price discrimination against nonwhites in ghetto areas in all three cities, though exclusionary practices have increased in suburban areas in Philadelphia.

A decade of change in drawings by black children.

Comparison of human figure drawings made by black children before 1960 with those made by black children after 1970 indicated that more black racial indicators are included in current drawings. These results suggest that black children are now more positively aware and accepting of their racial and ethnic background than was the case before 1960. The authors offer possible explanations.