USING ‘‘SOCIALLY ASSIGNED RACE’’ TO PROBE WHITE ADVANTAGES IN HEALTH STATUS Objectives: We explore the relationships be- Camara Phyllis Jones, MD, MPH, PhD; Benedict I. Truman, MD, MPH; tween socially assigned race (‘‘How do other LaurieD.Elam-Evans,PhD,MPH;CamilleA.Jones,MD,MPH; people usually classify you in this country?’’), self- identified race/ethnicity, and excellent or very ClaraY.Jones,MD,MPH;RuthJiles,PhD;SusanF.Rumisha,MSc; good general health status. We then take Geraldine S. Perry, DrPH advantage of subgroups which are discordant on self-identified race/ethnicity and socially assigned race to examine whether being classified by others We posit that ‘‘race’’ acts on health as White conveys an advantage in health status, INTRODUCTION even for those who do not self-identify as White. through race-associated differences in life Racial health disparities have been experiences and life opportunities in our Methods: Analyses were conducted using documented in the United States since race-conscious society. That is, we posit pooled data from the eight states that used data on ‘‘race’’ and health have been that ‘‘race’’ is a potent predictor of health the Reactions to Race module of the 2004 1–4 outcomes in this country because of Behavioral Risk Factor Surveillance System. jointly collected. The question re- mains, however, why the variable ‘‘race’’ racism, which Jones has defined as ‘‘a Results: The agreement of socially assigned is such a potent predictor of health system of structuring opportunity and race with self-identified race/ethnicity varied outcomes, especially when it is widely assigning value based on the social 12 across the racial/ethnic groups currently defined acknowledged that ‘‘race’’ is a social interpretation of how one looks.’’ by the United States government. Included construct, not a biological descriptor.5–9 Jones proposes that ‘‘race’’ be formally among those usually classified by others as understood as the social interpretation of White were 26.8% of those who self-identified We gain some insight into this question as Hispanic, 47.6% of those who self-identified by observing that the ‘‘race’’ noted by a our physical appearance in a given place as American Indian, and 59.5% of those who hospital admissions clerk on a medical and time, and she suggests that it can be self-identified with More than one race. record is the same ‘‘race’’ noted by a sales measured by a person’s response to the Among those who self-identified as Hispanic, clerk in a store, a taxi driver or police question ‘‘How do other people usually the age-, education-, and language-adjusted classify you in this country?’’12 Note that proportion reporting excellent or very good officer on the street, a judge in a court- 10–12 this ‘‘socially assigned race’’ is distinct health was 8.7 percentage points higher for room, or a teacher in a classroom, from self-identified race/ethnicity, and those socially assigned as White than for those and, in our opinion, this ‘‘race’’ is quickly socially assigned as Hispanic (P5.04); among could be a useful tool for probing the and routinely assigned without the benefit those who self-identified as American Indian, impacts of racism on health because it of queries about self-identification, ances- that proportion was 15.4 percentage points measures the ad hoc racial classification higher for those socially assigned as White than try, culture, or genetic endowment. upon which racism operates. for those socially assigned as American Indian Indeed, this ad hoc racial classification In this article, we explore the relation- (P5.05); and among those who self-identified hasbeenaninfluentialbasisforinterac- with More than one race, that proportion was ships between ‘‘socially assigned race,’’ self- tions between individuals and institutions 23.6 percentage points higher for those socially identified race/ethnicity, and excellent or in our society for centuries.13 assigned as White than for those socially very good general health status. We then assigned as Black (P,.01). On the other hand, take advantage of subgroups that are no significant differences were found between those socially assigned as White who self- discordant on self-identified race/ethnicity identified as White and those socially assigned From the Division of Adult and Com- and ‘‘socially assigned race’’ to examine as White who self-identified as Hispanic,as munity Health (CPJ, LDEE, RJ, GSP) and whether being socially assigned as White American Indian,orwithMore than one race. Office of Minority Health and Health Disparities (BIT), Centers for Disease Con- conveys an advantage in health status, even for those who do not self-identify as White. Conclusions: Being classified by others as White trol and Prevention, Atlanta, Georgia; Cin- is associated with large and statistically significant cinnati Health Department, Cincinnati, Using ‘‘socially assigned race’’ to probe advantages in health status, no matter how one Ohio (CAJ); Department of Public Health advantages in health status associated with self-identifies. (Ethn Dis. 2008;18:496-504) and Family Medicine, Tufts University being classified by others as White,weaim School of Medicine, Boston, Massachusetts to further elucidate the impacts of racism on (CYJ); National Institute for Medical Re- Key Words: Behavioral Risk Factor Surveil- health. lance System, Racism, Self-rated Health search, Dar es Salaam, Tanzania (SFR). Address correspondence and reprint requests to: Camara Phyllis Jones, MD, METHODS The findings and conclusions in this paper MPH, PhD; Centers for Disease Control are those of the authors and do not necessarily and Prevention; 4770 Buford Highway NE; represent the official position of the Centers Mailstop K-67; Atlanta, Georgia 30341; The Behavioral Risk Factor Surveil- for Disease Control and Prevention. (770) 488-5268; [email protected] lance System (BRFSS), developed by 496 Ethnicity & Disease, Volume 18, Autumn 2008 ‘‘SOCIALLY ASSIGNED RACE’’ AND HEALTH - Jones et al the Centers for Disease Control and question on race, their self-identified group but are socially assigned to the Prevention (CDC), is an ongoing state- race/ethnicity was coded as the racial White group, does their general health based system of health surveys admin- group they selected (White, Black or status differ from a) the health of those istered by telephone to a representative African American, Asian, Native Hawai- who both self-identify with and are sample of non-institutionalized persons ian or Other Pacific Islander, American socially assigned to the particular non- aged $18 years. Details on the objec- Indian or Alaska Native,orOther). If White group, and b) the health of those tives, design, use, and limitations of the respondents answered No to ‘‘Are you who both self-identify with and are BRFSS can be found elsewhere.14–16 Hispanic or Latino?’’ and selected more socially assigned to the White group? The Reactions to Race module is a six- than one racial group, their self-identi- Post-stratification weights were used question optional module first devel- fied race/ethnicity was coded as More to adjust for probability of selection and oped for the BRFSS in 2001 by the than one race. nonresponse.19,20 SAS version 8.2 (SAS CDC Measures of Racism Working The socially assigned race variable Institute, Inc., Cary, NC) with SU- Group.17 The questions include assess- was based on responses to the first DAAN version 9 (RTI International, ments of socially assigned race (‘‘How question asked on the BRFSS Reactions Research Triangle Park, NC) was used do other people usually classify you in to Race module: ‘‘How do other people for statistical analyses to account for the this country?’’) and race consciousness usually classify you in this country? complex sampling design. Comparisons (‘‘How often do you think about your Would you say White, Black or African of the outcome between subgroups race?’’), as well as perceptions of American, Hispanic or Latino, Asian, jointly defined by self-identified race/ differential treatment at work and when Native Hawaiian or Other Pacific Is- ethnicity and socially assigned race were seeking health care, and reports of lander, AmericanIndianorAlaska adjusted for reported age in years, emotional upset and physical symptoms Native,orSome Other Group?’’ Re- education level (none or kindergarten, as a result of race-based treatment. The sponse categories included all of the grades 1–8, grades 9–11, grade 12 or Reactions to Race module underwent federal Office of Management and GED, college 1 to 3 years, or college 4 three rounds of cognitive testing, one Budget (OMB) ‘‘race’’ categories as well or more years), and respondent prefer- round of field testing, and pilot testing as the OMB and ethnicity categories.18 ence for questionnaire language (En- by six invited states on the 2002 BRFSS. General health status was assessed glish or Spanish) using predicted mar- This article presents analyses of pooled using the self-rated health question from ginals from logistic regression models.21 data from the eight states (Arkansas, the BRFSS core questionnaire: ‘‘Would Differences were considered statistically Colorado, Delaware, District of Colum- you say that in general your health is significant at P#.05. bia, Mississippi, Rhode Island, South Excellent, Very good, Good, Fair, or Carolina, and Wisconsin) that used the Poor?’’ Response categories Excellent Reactions to Race module in 2004, the and Very good were combined in this RESULTS first year it was made available to all study to serve as a measure of optimal states. health, the outcome of interest, in Table 1 presents the joint distribu- The self-identified race/ethnicity contrast to response categories Good, tion of the 34,775 respondents in our variable was constructed from two Fair, and Poor, which do not represent sample by self-identified race/ethnicity separate questions included on the optimal health. Higher levels of Excel- and socially assigned race, as well as the BRFSS core questionnaire: ‘‘Are you lent or Very good health are considered weighted percent distribution of socially Hispanic or Latino? [Yes, No]’’ and an advantage in health status.