Issues in the Assessment of “Race” among Latinos: Implications for Research and Policy

Abstract

Measurement of race and ethnicity is integral to assessing and addressing health disparities experienced by minorities. However, the unique experiences of Latinos related to race and the discordance between understandings of race among Latinos and the predominant U.S. conceptualizations of this construct impact how Latinos respond to measurement approaches. As a result, data collection methodologies often yield ambiguous responses that reveal little about this population. This paper examines Latinos’ racial responding, and how this relates to their experiences and understanding of their racial identity. We recommend the use of a combined race and ethnicity question and open-ended race and ethnicity questions, when feasible, which will likely yield more meaningful data that can be used to address this populations’ health needs.

Identifying health disparities is predicated on the ability to measure racial and ethnic differences in health outcomes (Hasnain-Wynia & Baker, 2006; Moscou, Anderson, Kaplan, & Valencia, 2003). In addition to identifying disparities, data on race and ethnicity have uses ranging from assessing communities’ health needs to informing the creation of interventions (Buescher, Gizlice, & Jones-Vessey, 2005; Friedman, Cohen, Averbach, & Norton, 2000; Hasnain-Wynia & Baker, 2006; Lin & Kelsey, 2000; Lurie & Fremont, 2006; Nerenz, Hunt, & Escarce, 2006). However, discordant conceptualizations of race and ethnicity between policy makers, researchers, and the public present challenges in attempting to measure these constructs (Centers for Disease Control and Prevention, 1993; Moscou, et al., 2003). Such discordance has unique implications for people of Hispanic origin, and disparities in morbidity and mortality experienced by Latinos illuminate the need for methods that accurately characterize this population. However, current methodologies yield ambiguous data on “race,” as demonstrated by the selection of “some other race” by 42% of Latino respondents to Census 2000 (Grieco & Cassidy, 2001). The selection of this response is likely due to different conceptualizations of race and ethnicity among respondents compared to those designing the instrument, and may provide evidence for Latinos’ non-identification with the United States’ taxonomy of race and ethnicity.

Methodologies which view race and ethnicity as mutually exclusive, such as those used by the federal government, are inconsistent with the understanding of these constructs among Latinos (Lee & Tafoya, 2006; Tienda & Mitchell, 2006). Furthermore, such methodologies fail to account for the varied conceptualization of race and ethnicity present in Latin America and among many Latinos in the U.S. (Duany, 1998; Nutini, 1997; Rodriguez, Castro, Garcia, & Torres, 1991). In addition, the terms Hispanic and Latino were created by the U.S. government and have little meaning in Latin America (Tienda & Mitchell, 2006). This discordance between methodologies and respondents’ understanding of their identity may yield responses that have no significant meaning for individuals (Borrell & Crawford, 2005; Moscou, et al., 2003; Rodriguez, 1992). When data that have little to no meaning for respondents are collected, analyzed, and reported they can significantly alter our understanding of a population’s health needs, masking some disparities while overestimating others (Buescher, et al., 2005; Friedman, et al., 2000). Furthermore, the reclassification of “some other race” responses used by Latino respondents to a racial group (e.g. white) not only impacts our understanding of the health concerns of Latinos, but has implications for understanding the health concerns of other communities.

Identification with a racial and/or ethnic group is likely to result in an individual having beliefs, health practices, and risk factors characteristic of others belonging to that group (Hasnain-Wynia & Baker, 2006). Insight into this identification can prove useful in understanding the health of Latinos and developing culturally competent interventions. Therefore, a system of measurement that assesses race and ethnicity in a way that is meaningful for Latinos is necessary if health disparities are to be adequately identified, understood, and addressed. A revised method that reflects salient forms of identification has the potential to provide health researchers with a more accurate and meaningful characterization of Latino respondents. This will allow for a better understanding of populations and their health needs, advancing progress toward the elimination of disparities in health. This paper aims to examine Latinos’ identification, or lack thereof, with currently used racial and ethnic categories and will discuss the history of racial classification in the U.S., the potential for misrepresenting health disparities, the implications these issues have for research and policy, and the ways in which understanding Latino identity can be used to improve current methodologies.

Historical background

The Institute of Medicine defines race as “a sociocultural concept wherein groups of people sharing certain physical characteristics are treated differently based on stereotypical thinking, discriminatory institutions and social structures, a shared worldview, and social myths,” while ethnicity is defined as “shared culture and way of life…reflected in language, folkways, religious and other institutional forms” (Smedley, Stith, & Nelson, 2003). A historical examination of the federal classification of race among Latinos reveals varying approaches. In 1930, the U.S. Census Bureau first attempted to enumerate individuals of Hispanic origin by including “Mexican” as a race, although this was later removed due to concerns that Mexican was not a racial group (Prewitt, 2002). Subsequent censuses identified Latinos by respondents’ birthplace, Spanish as the native language, and Spanish surname, although Latinos were designated as “white” (Borrell & Crawford, 2005; Rodriguez, 1992; Rodriguez, et al., 1991; U.S. Bureau of Census, n.d.). Such methodologies were problematic due to potential under or over counting resulting from marriage, Filipino nationality, or being a non-Spanish speaking Latino (Aguirre-Molina, Molina, & Zambrana, 2001).

The 1970 census introduced the term “Hispanic” to refer to individuals of any race who have origins in Mexico, Puerto Rico, Cuba, Central or South America, or other Spanish cultures (Office of Management and Budget, 1997; Waterston, 2006). Also, for the first time respondents’ race and ethnicity were collected separately based on self-reported identification rather than enumerator observation (Gibson & Jung, 2002; Tienda & Mitchell, 2006). In Census 2000, “Latino” was added to refer to individuals of Latin American origin, such that the ethnicity question read “Is this person Spanish/Hispanic/Latino?” and the race question read “What is this person’s race?” (Grieco & Cassidy, 2001; Hayes-Bautista & Chapa, 1987; U.S. Bureau of Census, n.d.). Since 1970, the Office of Management and Budget (OMB) has mandated that federal data collection efforts separately measure respondents’ race and ethnicity when such data are collected through self-report (Office of Management and Budget, 1997; U.S. Bureau of the Census, 2000). This methodology was intended to provide a more detailed enumeration of the Latino population, in an effort to ensure that such communities were appropriately allocated federal services and programs (Tienda & Mitchell, 2006). However, for many Latinos the differentiation between race and ethnicity is inconsistent with their lived identities (Lee & Tafoya, 2006; Tienda & Mitchell, 2006). Given the central role of these constructs in assessing health disparities, it is imperative that the measurement of race and ethnicity be consistent with respondents’ understanding (Centers for Disease Control and Prevention, 1993). Otherwise, conclusions drawn about racial differences in health outcomes may be questionable due to the methodology not accurately assessing the racial group, or groups, with which respondents identify.

Congruence between measurement and lived identity

The social conceptualization of race in the United States bases race on a person’s physical appearance and ancestry and is generally dichotomized between white and black, with other racial groups fitting “in between” (Landale & Oropesa, 2002; Rodriguez & Cordero-Guzman, 1992). This differs greatly from the Latin American understanding which includes variables such as birthplace, culture, and social class, and is not simply one’s physical appearance and ancestry (Duany, 1998; Nutini, 1997; Rodriguez, 1992). Such a varied conceptualization has resulted in a Latin American racial taxonomy that may differ in meaning according to context, and that takes into account phenotypic diversity among Latinos (Duany, 1998; Nutini, 1997; Rodriguez, et al., 1991). Latinos familiar with this understanding of race may experience difficulty finding their place in the U.S. racial taxonomy and may not identify with the system, instead identifying in a way that is reflective of their experiences with their identity. This lack of fit with common methods of assessing race leads many Latinos to perceive their identity as being outside of the standard race response options. For example, in one recent study conducted with Spanish-speaking immigrants, 75% of respondents who identified their ethnicity as Hispanic/Latino classified their race as “other” (Kaphingst, Lachance, Gepp, D’Anna, & Rios-Ellis, 2010). Rodriguez and Cordero-Guzman (1992) examined why the majority of a sample of 58 Latinos indicated their race as “other.” Over half (63.5%) of these respondents cited cultural reasons, including family and culture (i.e. having “roots” in a Latin American country), one’s birthplace (i.e. being born in a Latin American country), one’s personal politics (i.e. not identifying with white Americans), and being raised with a Latino identity. For many, selecting “other” indicated that race is a variable concept that is not determined solely by one’s ancestry or appearance (Rodriguez, 1992).

Latinos’ non-identification with the U.S. racial taxonomy may result from experiences with discrimination and oppression. Holley, Salas, Marsiglia, Yabiku, Fitzharris, and Jackson (2009) found that experiencing discrimination caused Mexican adolescents to have a solely Mexican “racial” identity, rejecting the U.S. racial taxonomy and constructing an identity that was influenced by racial discrimination. Furthermore, the authors suggested that possessing a solely Mexican identity, without coupling it with a racial identity (e.g. white), may be a form of resistance to oppression. This was thought to be especially true among Mexican adolescents who may deal with racist stereotyping (Holley, et al., 2009). Similarly, Golash-Boza and Darity (2008) in their analysis of the 1989 Latino National Political Survey (LNPS) and the 2002 National Survey of Latinos (NSL) found that respondents who experienced discrimination were less likely to identify as white than those who did not have such experiences, and more likely to identify as black, other, or Hispanic. Both Holley, et al. (2009) and Golash-Boza and Darity (2008) concluded that experiences with racial discrimination leads some Latinos to perceive that they are not viewed as white within the U.S. racial taxonomy, particularly when the perpetrators are white. Such experiences may then lead to a non-white racial identification among Latinos, especially among dark-skinned Latinos who reported experiencing discrimination more than their light-skinned counterparts (Golash-Boza & Darity, 2008). While this non-white identity may fall within the spectrum of the U.S. taxonomy of race, often it does not.

Latinos who view the racial classification system as non-inclusive of their identity may reject it altogether, developing a panethnic identity that prioritizes one’s ethnic identity as Latino over a “racialized” classification (Landale & Oropesa, 2002). Panethnicity is “the expansion of ethnic group boundaries to include different national or ethnic groups that share a common language, a common culture, or a common regional origin into an encompassing identity” (Itzigsohn & Dore-Cabral, 2000). For example, in a study conducted by Itzigsohn and Dore-Cabral (2000) that examined self-identification among Dominican immigrants through open-ended questions, respondents asked their race in the Dominican Republic predominately used the racial identifiers black (38.3%), indio (40%) or white (6.7%). However, when asked their race in the U.S., the majority of these same respondents used panethnic identifiers such as Hispano (46.7%) and Latino (16.7%). Additionally, a study by Vaquera and Kao (2006) found that the development of a panethnic identity resulted in 67% of Hispanic adolescents selecting their race as “other” or leaving the question blank, illustrating the lack of relevance of the U.S. racial classification system to the lived realities of Latino identity (Vaquera & Kao, 2006).

The panethnic terms Hispanic and Latino are uniquely American creations that have no real meaning in Latin America (Tienda & Mitchell, 2006). Therefore, panethnic identification through the use of these terms becomes salient as a result of immigration to the U.S. In support of this, there is a relationship between time in the U.S. and racial identification. In the study by Itzigsohn and Dore-Cabral (2000), participants who identified panethnically had been in the U.S. significantly longer (12.5 years) than those who identified alternatively (8 years). Racial identification is also influenced by how far removed one is from the immigrant generation, but data have not reflected a consistent trend over time. Golash-Boza and Darity (2008) examined this issue and identified contradictory findings between the 1989 LNPS and the 2002 NLS. The 1989 LNPS revealed an increase in panethnic identification from the first generation of immigrants to the third generation of U.S.-born Latinos. Conversely, the 2002 NLS demonstrated a moderate decline in panethnic identification from first generation immigrants to fourth generation U.S.-born Latinos.

The influence of length of time in the U.S. and generation on Latinos’ racial identity is related to acculturation, the modification of one’s culture resulting from contact with another (Smedley, et al., 2003). Vaquera and Kao (2006) found that Latinos’ acculturation into American society led to greater identification with the U.S. racial taxonomy; specifically among third generation Latino adolescents who may be more familiar with the racial system and may feel the need to identify accordingly. Conversely, acculturation strengthened the preference for panethnic identification in the study by Itzigsohn and Dore-Cabral (2000), who suggested that acculturation into American society may result in acceptance of the idea that people of Latin American descent are viewed panethnically, leading to the formation of a salient panethnic identity. These contradictory findings, like those on the impact of generation, illuminate an area in need of greater examination. Furthermore, given that almost half (48%) of the population in the Vaquera and Kao (2006) study was Mexican and a third (33.4%) Central/South American/other, while the population in the Itzigsohn and Dore-Cabral (2000) study was Dominican, these conflicting findings may reveal the diversity of experiences had by Latinos of varied national origins when dealing with issues of race.

Adding to the complexity of this issue is the fact that although current federal standards classify Latino as an ethnic identity (Office of Management and Budget, 1997), Latinos are often treated as a racial group in U.S. society (Itzigsohn & Dore-Cabral, 2000; Landale & Oropesa, 2002; Rodriguez, et al., 1991; Tienda & Mitchell, 2006). Various experiences, such as the reporting of health statistics for Latinos alongside those of other racial groups and experiences of discrimination based on Latino identity, often result in the public understanding of Latino as a race (Golash-Boza & Darity, 2008; Tienda & Mitchell, 2006). This lack of congruence between methodology and respondents’ subjective understanding of their identity may elicit responses to questions of race that do not speak to the respondents’ lived identity and that have no significant meaning for the individual (Borrell & Crawford, 2005; Moscou, et al., 2003; Rodriguez, 1992). A qualitative study conducted by Rodriguez (1991) found individuals’ responses concerning race might not reflect a salient personal identification with a racial group, and may be secondary to their Hispanic identity or their country of origin identity.

When data on race are reported and interpreted, the implication is often that this has some significance for respondents. While racial responses may be significant for some, for others this has little meaning beyond identifying the color of one’s skin (Rodriguez, et al., 1991). The underlying assumption of racial group membership is not relevant in such cases. Therefore, generalizations and conclusions about the health status of Latinos, specifically Latino subgroups, based on racial group membership may be invalid and subsequent efforts to address disparities based on such data may be ineffectual.

Reclassification of responses

An additional way in which lack of congruence between methodologies and individuals’ understanding of their identities can impact public health surveillance is through reclassification of “some other race” responses. The issue of reclassification is of particular significance to the understanding of Latinos who accounted for 97% of all “some other race” responses in Census 2000 (Grieco & Cassidy, 2001). Buescher, Gizlice, and Jones-Vessey (2005) examined the impact of reclassification on racial disparities among women in North Carolina using birth certificate data from 2002. Of the 15,704 birth records for which the mother self-identified as Hispanic, 10,361 indicated their race as “other” with the great majority (9,445) providing “Hispanic” as their write-in response. Consistent with the National Center for Health Statistics (NCHS) standards for coding birth record data, this group of responses was reclassified as “white”(National Center for Health Statistics, 1999). This reclassification not only increased the white population in the total sample from 63.4% to 72.7%, but also significantly changed the understanding of disparities in health outcomes. Among whites the rate of maternal smoking during pregnancy decreased from 15.9% to 14.1% and the percentage of women with late or no prenatal care increased from 10.1% to 12.5%.

Similarly, Friedman, Cohen, Averbach, and Norton (2000) examined 1997 birth records in Massachusetts and found that 96% of Dominicans and 64% of Cape Verdeans self-identified as “other.” When these responses were reclassified as “black,” as mandated by NCHS, the statewide infant mortality rate for blacks decreased from 11.1 deaths per 1000 live births to 10.4 per 1000. These findings demonstrate that reclassifying “other” responses impacts the characterization of Latino respondents and can significantly alter the understanding of disparities across racial and ethnic groups. These results also provide additional evidence regarding how understanding of health disparities may change when racial measurement elicits responses that are not salient to those being measured.

Measurement methodology

Although measurements of race and ethnicity are intended to assess the identification of respondents, responses to questions of race are not solely influenced by respondents’ subjective understanding of their identity. Prior research indicates that there are methodological factors that also influence responses. Hirshman, Alba, and Farley (2000) examined answers to varied question formats and found that when separate race and ethnicity questions were asked, 66.4% of Hispanics identified their race as white. However, when presented with a combined race and ethnicity question, identification as white was reduced to 13.7% and 56.4% of participants identified as Hispanic only. Other studies have also found that the majority of respondents to a combined question identified as solely Latino (Campbell & Rogalin, 2006), even when having the option of selecting multiple races.

Additional factors such as the presence or absence of the “some other race” response option and the order in which race and ethnicity questions are presented also impact Latinos’ racial responding (Hirschman, Alba, & Farley, 2000; Hitlin, Brown, & Elder, 2007). Although the OMB mandates that when race and ethnicity are measured separately ethnicity is to be measured first, these findings are still relevant in that they demonstrate the instability of racial responses resulting from minor methodological changes. Interestingly, there was discussion of eliminating the “some other race” response option from Census 2010 (U.S. Bureau of Census, 2003). Although it was decided that the response option will remain, a revised system that reflects the lived identities of respondents may be a more favorable alternative for future data collection efforts than eliminating a category widely used by Latino respondents.

Implications for research and policy

A combined race and ethnicity question may allow respondents to report their identity in a way that is most consistent with their everyday experiences and understanding of these constructs. While separate questions may “force” a response even when the selection has no significant meaning for the respondent, a combined question encourages respondents to consider the identity that they find most salient (Campbell & Rogalin, 2006). A combined question is also beneficial in that it can eliminate the use of the terms “race” and “ethnicity,” which are potentially confusing and vary in meaning. Such a question could simply state, “Is this person…White, Black/African American, American Indian or Alaska Native, Asian, Native Hawaiian or Other Pacific Islander, Spanish/Hispanic/Latino, Some other race” similar to the format used in the 1980 Census, allowing respondents to indicate the response option, or multiple options, with which they identify (Lee & Tafoya, 2006; U.S. Bureau of the Census, n.d.).

Another potentially important alternative for public health research is the collection of information on ancestry, country of origin, birthplace, and open-ended data on racial and ethnic identification (Baker, et al., 2006). This can provide health institutions with a thorough understanding of the communities they serve, particularly in areas with large immigrant populations for whom assessing country of origin may be more meaningful than assessing race, which is both ambiguous and non descriptive of reality for many Latinos (Buescher, et al., 2005). Furthermore, given the salience of ethnic identification among immigrant as well as U.S.-born Latinos, the collection of information on Latino sub-groups (e.g. Mexican) may provide a more refined characterization of this diverse population. This would allow Latinos to continue to indicate the sub-groups with which they identify, as is currently done, if a combined race and ethnicity question is used. In addition, exploring disparities between Latino sub-groups may reveal health concerns among specific populations. Such detailed demographics can then be used to target the needs of particular communities, neighborhoods and individuals (Hasnain-Wynia & Baker, 2006).

National data collection efforts may find the use of additional questions, particularly open-ended questions, cumbersome. However, state and local efforts may find additional race and ethnicity questions feasible and beneficial in providing detail on health differences among local racial and ethnic groups (Friedman, et al., 2000; Hasnain-Wynia & Baker, 2006). Open-ended data could then be aggregated to the standard OMB categories, when needed, to provide greater comparability between local, state, and federal data on health differences by race and ethnicity (Hasnain-Wynia & Baker, 2006). Data collection approaches that can assess these constructs among individuals with varying levels of literacy skills is critical, particularly for Latino populations, who have lower average levels of literacy than other racial and ethnic groups in the U.S. (Kutner, Greenberg, & Baer, 2003).