Page 81 - IJPS-2-2

Haiyan Zhu, Qiushi Feng, and Danan Gu

1997; Idler, Hudson, and Leventhal, 1999). SRH is widely used in health and mortality studies due
to its simplicity in collecting data and its effectiveness in evaluating overall health status and pre-
dicting health outcomes, health care utilization, and subsequent mortality (e.g., Anson, Shteingrad,
and Paran, 2011; Chen and Wu, 2008; Ferraro, Farmer, and Wybraniec, 1997; Idler and Benyamini,
1997). However, issues remain despite SRH’s widespread use (Smith and Goldman, 2011). Because
the complex process of self-rating is inherently subjective, sometimes the self-rating may be biased
(Huisman and Deeg, 2010; Jylhä, 2009). Accordingly, the subjective nature of SRH may affect the
accuracy of the overall health assessment and the ability of that assessment to predict mortality.
In contrast, another line of literature has long highlighted the advantages of external health ratings.
Most studies in this field focused on medical ratings by a professional (e.g., Glare, Virik, Jones et al.,
2003; Rocker, Cook, Sjokvist et al., 2004). Beyond the professional assessment, an early study by
van Doorn (1998) suggested that a health report from a spouse could also predict mortality of his/her
partner, independent of SRH and many other objective health measures. This finding on spouse’s
health report gained confirmation from some subsequent studies (Daugherty, 2009; Peek, Stimpson,
Townsend et al., 2006). Meanwhile, a new line of research started to emerge with a focus on health
evaluation by survey interviewers (Chen and Wu, 2008; Feng, Zhu, Zhen et al., 2016; Smith and
Goldman, 2011; Todd and Goldman, 2013). According to these studies, the use of interviewer-rated
health (IRH) may have some advantages over SRH in capturing health situations and may comple-
ment the routine use of SRH. For example, unlike SRH, IRH avoids person-specific biases from the
respondent, applies a sound comparative framework for judgment based on multiple respondents,
incorporates good on-site observations about living conditions and environment, and often takes ad-
vantage of the evaluation of SRH from the respondent (Brissette, Leventhal, and Leventhal, 2003;
Feng, Zhu, Zhen et al., 2016).
Among the pioneering studies investigating SRH and IRH as tools for mortality prediction, there
have been few efforts to examine their relative predictive power across different subpopulations. For
example, in a recent study, Feng et al. (2016) examined the potential of IRH as a complementary
measure to SRH by comparing their components and predictive powers for mortality in the Chinese
elderly population. Analyzing data from a nationwide survey, they found that SRH and IRH captured
similar health information, but SRH placed more weight on health perceptions and experiences while
IRH emphasized more objective health conditions such as IADL (instrumental activities of daily
living) and ADL (activities of daily living) disabilities. Importantly, this study showed that IRH was
a strong predictor of mortality, independent of SRH, and thus IRH could be used as a good measure
to complement SRH among Chinese elders. However, this study did not examine the predictive
powers of SRH and IRH in mortality among different subpopulations of Chinese elders, so it is un-
known whether these established findings would hold across specific population groups, such as dif-
ferent demographic and socioeconomic groups.
SRH may predict mortality differently across subpopulations. Due to the subjective nature of SRH,
individual characteristics including age, gender, and education can affect the understanding and
judgment of one’s health status and the referents and criteria used to assess overall health (Dowad
and Zajacova, 2007; 2010; McFadden, Luben, Bingham et al., 2009). Individuals with different
backgrounds may therefore weigh and value the SRH domains differently and may also use different
referents when evaluating their global health. For example, SRH and its predictive power for mortal-
ity may vary by socioeconomic status (SES). Research has shown that individuals with lower levels
of educational attainment likely place more weight on health behaviors (Krause and Jay, 1994);
SRH could be a more accurate predictor of mortality for people with higher socioeconomic sta-
tus because they may have a better understanding of their personal health due to the better health
resources compared to those with lower educational attainment and income (Franks, Gold, and
Fiscella, 2003; Quesnel-Vallee, 2007). In addition, rural/urban residence, which is often considered
as an indicator of SES (Zhu and Xie, 2007), may also matter, especially for countries such as China

International Journal of Population Studies | 2016, Volume 2, Issue 2 75