Expanding Discrimination Research: Beyond Ethnicity and to the Web^*

Abstract

Objective. This article aims to expand research about perceptions of discrimination both substantively and methodologically beyond the domains of race and ethnicity, relying partly on web-based surveys. Methods. We conducted parallel surveys over the telephone and the World-Wide Web, using standard random-digit dial (RDD) techniques for the former, and a large volunteer panel for the latter. Results. Both modes, phone and web, revealed that respondents consider discrimination based on physical appearance and economic status to be more prevalent than discrimination based on ethnicity. Respondents also reported that they themselves have been victimized more by physical appearance and economic-status discrimination than by ethnic discrimination. Significant differences emerged between the phone and web respondent pools, even after controlling for such independent variables as age, race, education level, and gender. Conclusions. People perceive discrimination across many aspects of social life, and appear more willing to reveal knowledge about controversial social phenomena on the web than on the phone.

Sex Differences in the Acceptability of Discrimination

“…In general, we expected to find less tolerance for discrimination on the telephone than on the Web, except perhaps on the Arab American question, where respondents might consider it "socially desirable" to express a prodiscrimination position to a live interviewer. This hypothesis is consistent with other findings about the perceived prevalence of discrimination (Kuran and McCaffery 2004;Krysan 1998). We also expected to encounter interactions between mode and sex, with men exhibiting greater acceptance of discrimination on the Web than on the telephone, because of the social desirability bias.…”

“…In the Web survey, the sex gap was substantially larger. It thus appears, in line with other recent survey research (Holbrook and Krosnick 2005;Kuran and McCaffery 2004;Dillman et al 2003;Holbrook, Green, and Krosnick 2003;Aquilino 1994), that a social desirability bias makes expressed attitudes vary according to the anonymity granted to respondents. More intriguing, the social desirability bias induced by live interviewers may operate in opposing directions for men and women, with men relatively understating and women relatively overstating their tolerance for discrimination, causing a convergence of attitudes.…”

CEO pulchronomics and appearance discrimination

“…Altonji and Blank, 1999). In a survey conducted over the telephone and the Internet, respondents consider discrimination based on looks to be more prevalent than discrimination based on ethnicity (Kuran and McCaffery, 2004). Our empirical evidence of appearance discrimination not only supports the sentiment expressed in this survey but also adds a new dimension to the existing discrimination literature largely focused on race, age and gender.…”

Typology of Risks of Generative Text-to-Image Models

“…Bias investigations in language technologies (as in the social sciences [91,150]) have typically centered on a narrow range of salient demographics, possibly underestimating the full extent of discrimination [21,46,66] . In line with the findings from NLP research [21], there is a primary focus on dataset bias, with other sources of bias in the model life cycle being underexplored.…”