Crossracial Differences in the Racial Preferences of Potential Dating Partners: A Test of the Alienation of African Americans and Social Dominance Orientation

Abstract

Studying interracial romance has been useful for understanding general race relations. Theories of African American alienation and social dominance orientation help explain why previous research has found African Americans to be the least desired racial dating partners. Alienation predicts that African Americans are less willing to interracially date than other racial groups since they are not allowed to participate in the majority culture. Social dominance orientation predicts that African Americans are more willing to interracially date than other racial groups because they occupy the lowest position in our racial hierarchy. This study utilizes an Internet dating website to explore the racial dating preferences of European Americans, African Americans, Hispanic Americans, and Asian Americans. The theory of African American alienation is upheld, as African Americans are generally less willing to interracially date than other races and are especially less willing to date European Americans.

ACKNOWLEDGMENTS

I thank Nicole Dash, Kevin Yoder, Starita Smith, and the anonymous reviewers for their comments on this manuscript. I also thank Deborah Cosimo, Elizabeth Fisher, and Ashley Jarvis for helping me with the data coding.

NOTES

Notes

¹ It should be noted that recently, scholars have challenged Ogbu's ideas concerning the effects of oppositional culture on educational attainment (Flores-Gonzalez 2005; Foley 2005; Harris 2006).

² In fact, African Americans are about as likely to attend multiracial congregations as first-generation immigrants (Emerson 2006). This suggests that they may be as estranged from the larger society as those who are brand new to our culture, despite having been in the United States for the past few centuries.

³ Bonilla-Silva (2004) argues for a triracial, noncontinuous stratification structure that places whites as the highest group, honorary whites as the middle group, and collective blacks as the lowest group. Since he describes the movement between these groups as potentially fluid and there is a strong tendency toward maintaining the bottom racial group, it is unclear whether this triracial stratification theory postulates that blacks face a type of discrimination that differs in kind and degree from nonblacks. Thus, it is ambiguous whether this current research can document the accuracy of triracial stratification as a way to understand interracial romance.

⁴ For example, 98 percent of all public schools (Cattagni and Farris 2001) and 95.7 percent of all public libraries (Bertot and McClure 2000) had access to the Internet in 2000.

⁵ Babbie (1995) suggests that stratified sampling can be even more useful than even simple random samples since it allows the researcher to focus upon the dimensions the researcher has identified as particularly important.

⁶ Since the data would be weighted, I did not worry about finding a proper number of respondents to be chosen at each level of city size. Thus, the number of individuals chosen from each size of city was arbitrary. However, I choose more individuals from larger cities since it was easier to find enough individuals of a given minority race in those cities.

⁷ This occurred seven times during the collection of the data.

⁸ Interracial was created for all individuals who self-defined themselves as multiracial. It is impossible to make any genetic or phenotypic assessments of who placed him-/herself in this category.

⁹ Age may not be best specified as a continuous variable and a four-level education variable is also problematic. I tested to see if age is mis-specified with models with age as a logged variable. Such models neither increased the predictive abilities of the models nor was the logged age variable any more likely to be significant. I am limited to only four categories of education because of the way this variable was constructed on the Yahoo! website. However, I also tested education with a dummy variable based on whether the respondent had received a bachelor's degree or not. Once again, this variable neither increased the predictive abilities of the models nor was it any more likely to be significant in these models than the four level education variable.

¹⁰ Ideally, I would use a national sample that only included those seeking romantic partners. But such a sample does not exist. Even using just the unmarried population in the Census would provide a distorted number as many of those unmarrieds are either currently in a romantic relationship or are not seeking a romantic relationship. Thus, to simplify the assessment, I included all individuals in the census as a proxy for the distribution of the population in the United States.

¹¹ Using a weight procedure to link a nonprobability sample to a probability sample is recommended by Fowler (2002).

¹² When the oversample is excluded, 48.6 percent of the sample comes from large cities and 31.1 percent of the sample comes from medium cities. Dividing the 48.6 percent in large cities by 2.5 reduces the impact of all of those from large cities to 19.44 percent of the population, and dividing the 31.1 percent in medium cities by 1.8 reduces the impact of all those from medium cities to 17.27 percent of the population. This equalizes the impact of people from each size of city despite the higher numbers selected from the larger cities. When the oversample is added, then 47.5 percent of the sample comes from large cities and 32.7 percent of the sample comes from medium cities. The weight numbers of 5.6, 4.2, and 2.6 reduces the impact from large, medium, and small cities, respectively, to 8.48, 7.79, and 7.58 percent of the population. Such reductions make comparison of the percentages in Table 1 more accurate.

¹³ Compared with the findings of Fiebert et al. (2000), the European American respondents in this sample were less likely to date outside of their race while Asian American respondents were more likely to date outside their race. These differences are likely connected to the fact that Fiebert et al.'s research is based on a local sample in California and collected through written surveys, which are more susceptible to social desirability effects.

¹⁴ In the “Date White” model, Asian Americans are the reference group.

¹⁵ Because of the low number of respondents in these models, I calculated events per predictor variables (EPV) for each model. In models with Hispanics, this resulted in a score of 10.3, which is higher than the score of 10 that is recommended under most logistical models (Peduzzi et al. 1996). In Asian models, the EPV is only 8, but Vittinghoff and McCulloch (2007) suggest EPV scores in the range 5 to 9 do not usually generate problematic analysis.

¹⁶ In separate models for each racial group, I also tested education as a dummy variable separating those who received a college degree from those who had not. Education was not found to be significant in any of those models except for Asians asked about whether they would date blacks. In that model education was positively related to the propensity of Asians to be willing to date African Americans.