Page 15 - IJPS-5-1

Van Natta M and Zimmer Z

Hispanic citizens who were born in the U.S. are more likely than other groups to rate their health unfavorably. Finally,
and in contrast to our expectations, duration in the U.S. does not mitigate Spanish-speakers’ greater likelihood of rating
their health unfavorably. On balance then, in answer to our research questions, language does affect SAH, the association
between language and SAH is not substantially mediated by citizenship status as results hold when entering citizenship
into regression models, and it holds as well when considering ethnic variation.
These analyses are subject to a variety of limitations that could be addressed through future research. One important
limitation is the dichotomous nature of NHANES citizenship categories. Our analysis teased out a third category
(naturalized citizens) from available data, but information on permanent residence, visa-holder, and undocumented
statuses would improve understanding of the citizenship-health association. Pourat, Wallace, Hadler et al. (2014) have
developed a method of imputing undocumented status using answers to questions about permanent residence and length
of time in the U.S., but without such information in NHANES, such imputation is impossible. Future data collection
instruments would benefit from more variety in citizenship types, but they do so at the risk of alienating potential survey
participants who fear disclosing their legal status – particularly under the current U.S. administration (Wang, 2018).
Another limitation is NHANES’ categorization of ethnicity. There is no distinction between “ethnicity” and “race” in
this measure. This conflation particularly affects Spanish speakers, who although classified as “Hispanic” may self-identify
as any racial category (e.g., white or black) or multiple racial categories (Vega, Rodriguez and Gruskin, 2009). Furthermore,
ethnic categories seem to have been determined colloquially rather than critically. In the original NHANES categories,
“Mexican American,” for example, is not disaggregated between Mexican and Mexican American, despite distinct cultural,
and geopolitical meanings. Finally, collapsing multiple ethnicities into “other/multi-racial” sacrifices information, although
overall there were very few respondents in this category. Despite these drawbacks, these categorizations reflect important
classificatory trends in the U.S., where non-white residents face racial discrimination, tend to occupy lower socioeconomic
positions, and report worse SAH than their white counterparts (Ren, Amick, and Williams, 1999; Stuber, Galea, Ahern
et al., 2003; Cummings and Jackson, 2008). Future research should allow for a greater variety of ethnicities.
Finally, we recognize that many Hispanic respondents may speak both English and Spanish and that dichotomizing
respondents according to the language in which they took the survey may not fully capture the nuance of their bilingual
SAH interpretation. The so-called “acculturation” portion of the NHANES includes questions about language spoken at
home, as a child, with friends, and while reading, speaking, and thinking, but the utility of these questions for our analysis
is limited. Very few people actually responded to these questions, and they are not comprehensive enough to model
bilingualism in the kinds of equations we have employed in this analysis. Future research should focus on collecting more
robust information on bilingualism and how to model it adequately in relation to constructs such as SAH.
In spite of these limitations, our study has addressed an important issue regarding language and health status. Our
analysis supports recent studies (Bzostek, Goldman, and Pebley 2007; Viruell-Fuentes, Morenoff, Williams et al., 2011;
Sanchez and Vargas, 2016) in indicating that Spanish speakers are much more likely to rate their health as fair than
are others and that this is likely a function of the way in which the ubiquitous SAH item is translated in surveys. Our
models followed the examples of Bzostek, Goldman, and Pebley (2007) to include citizenship status and Viruell-Fuentes,
Morenoff, Williams et al. (2011) to treat SAH as multinomial. We advanced on these previous approaches using a large
nationally-representative and more recent dataset (NHANES 2003-2016) that included thousands of observations on
the variables of interest. We also include duration in the United States to account for acculturation. Given the likelihood
that the commonly used translation of the five SAH categories itself shapes the distribution of SAH responses, alternate
translations, such as replacing “regular” with “más o menos” to convey “fair” health (Viruell-Fuentes, Morenoff, and
Williams et al., 2011; Sanchez and Vargas, 2016), should be implemented in future survey design. Such validity questions
must be addressed if SAH is to be translated and employed across linguistic groups.
Other findings address issues that are so far underdeveloped in literature. Despite numerous barriers to health access
on the basis of legal status (Castañeda, Holmes, Madrigal et al., 2014; Martinez, Wu, Sandfort et al., 2015), our analysis
in fact finds that in some cases being a non-citizen or naturalized citizen increases one’s odds of a favorable health rating.
The positive contribution of the “healthy immigrant effect” is outweighed by the effect of being racialized as non-white
in the U.S., however, which associates with a less favorable SAH rating.
From a policy perspective, these results present a complex picture of immigrant health. Our study underscores the
need to critically examine interrelated factors of survey language, citizenship status, and ethnicity to understand the health
context of a large part of the U.S. population. It is crucial that large national level surveys of health meant to inform health-
care policy consider novel measures that overcome challenges faced in assessing the health of migrating, heterogeneous
populations. In particular, survey language may be a problem at the level of survey data collection and analysis, and the
effects of citizenship status should be considered alongside ethnicity to capture the role that ethnic minority status has in

International Journal of Population Studies | 2019, Volume 5, Issue 1 9