China Policy Journal Volume 1, Number 1, Fall 2018 | Page 101

China Policy Journal
statistically significant. In addition, we
find that age, education, and income of
respondents are significant antecedents
of subjective correlation, while gender
is not significant.
We graphically plot the relationship
between subjective and objective
air quality measures in Figure 1. The
distribution of sampled cities in the
plot implies that objective air pollution
is negatively related to subjective
air quality. For instance, Haikou emitted
least SO 2
among 32 sampled cities
(0.007 mg/m 3 ), and its subjective
perception was also highest (8.06 out
of 10). In contrast, the midyear emission
of SO 2
in Taiyuan was among the
highest group (0.062 mg/m 3 ) and its
subjective air quality was relatively
low (5.02). There are also some outliers,
however, owing to idiosyncratic
geographical and socioeconomic
characteristics. Coastal Qingdao had
a slightly higher concentration of SO 2
(0.059) than inland Lanzhou (0.052),
but Qingdao was top in subjective air
quality (7.644) while Lanzhou is at
the bottom (4.103), mostly because of
Qingdao’ coastal advantage to quickly
disperse air pollutants. Another potential
explanation is that subjective
air quality measures are influenced by
multiple sources of air pollutants. Although
Qingdao and Lanzhou were
similar in the concentration of SO 2
and
NO 2
, Qingdao’ PM 10
emission (0.092
mg/m 3 ) was much lower than Lanzhou
Figure 1. The relationship between objective and subjective air quality measures
98