Page 92 - IJPS-2-2

Self-rated health and interviewer-rated health: differentials in predictive power for mortality among subgroups of Chinese elders

merit of IRH across most of the demographic and SES subgroups, and therefore further expands the
previous literature by concluding that IRH can be a good complementary measure of SRH and
should be used in health and mortality studies (Feng, Zhu, Zhen et al., 2016; Smith and Goldman,
2011; Todd and Smith, 2013).
We also found that neither SRH nor IRH was associated with mortality in a few subgroups of
women, including those who were married, educated for at least one year, or had good family eco-
nomic conditions. The literature on how SES affects SRH’s predictive power is less than consistent.
Although some literature found that the associations between SRH and mortality were stronger
among higher SES groups of older Americans (Dowad and Zajacova, 2007; 2010), other research
has shown that the association between SRH and mortality was weaker among high-SES groups than
among low-SES groups (Singh-Manoux, Dugravot, Shipley et al., 2007). It is possible that the
low-SES women may not report serious illnesses because they cannot afford medical services, while
the high-SES groups may tend to over-report less serious health problems (Singh-Manoux, Dugravot,
Shipley et al., 2007). This pattern may be because the low-SES groups cannot afford medical ser-
vices but the high-SES groups have better health literacy and make more frequent visits to doctors
(Blackwell, Martinez, Gentleman et al., 2009). There is also evidence that women are more likely to
exaggerate minor health problems or report health problems at an earlier stage (Singh-Manoux,
Guéguen, Ferrie et al., 2008). As the SES of women in old Chinese cohorts is much lower than that
of men, it is possible that the higher SES women in our sample were more likely to overreport their
health problems than their low-status counterparts. Our additional analysis further revealed that as-
sociations between SRH/IRH and mortality were indeed explained away by physical and psycholog-
ical health conditions. A few reasons seem plausible here. Firstly, it is likely that women who are
married or have a high SES tend to assess their own overall health mainly based on ADL, IADL, cog-
nitive function, and chronic disease conditions. As a result, the significance of SRH would be ex-
plained by these health conditions. Secondly, when they provided health information to the inter-
viewer during interview, these higher SES women may have stressed these health conditions, and the
interviewers may thus have largely relied on these conditions, so that IRH also loses additional pre-
dictive power in the presence of these health conditions. Thirdly, the relatively small sample size and
lower mortality among married or higher position women compared to other subgroups may al-
so contribute to non-significant results. We call for more research to examine these issues.
There are some limitations to the study’s measurement of variables. Firstly, IRH is also a subjec-
tive measure because it depends on interviewers’ judgments (Feng, Zhu, Zhen et al., 2016). There-
fore, interviewers’ own characteristics could possibly affect how they rate respondents’ health status.
We call for future studies to measure and analyze interviewers’ characteristics, such as demograph-
ic characteristics, education, occupation, and health literacy, which could help improve the under-
standing of IRH. Secondly, the Chinese wording of the SRH and IRH categories is not exactly the
same, which may introduce some bias to respondents’ responses. However, since they largely repre-
sent the same meanings in Chinese, the inconsistency of wording should not seriously affect our
findings. In spite of these limitations, based on a national survey with a large sample size, we were
able to examine the predictive power of IRH relative to SRH for mortality in various subgroups of
the elderly Chinese population and found that, overall, IRH is an independent and robust predictor of
mortality and even performs better than SRH in mortality prediction among most of the subgroups.
5. Conclusions

Our study found that, in general, IRH is a robust predictor of mortality, independent of SRH across
most major demographic and socioeconomic subpopulations among Chinese older adults. Thus,
IRH could be a good complementary measurement for SRH. Among the subgroups of women in
which neither SRH nor IRH was significantly associated with mortality, we found that SRH’s and
IRH’s predictive power for mortality were explained away by respondents’ self-reported physical,

86 International Journal of Population Studies | 2016, Volume 2, Issue 2