Rebecca L. Brunson and Dr. Mark Showalter, Economics
In 1991 Dr. C Arden Pope published a study on the effect of changes in fine particulate pollution (PM10) on respiratory health, as measured by peak expiratory flow (PEF). The study found that respiratory health (PEF) was affected by changes in PM10 even when the levels were well below the national ambient air quality standard. The results of the study were statistically significant. “Elevated PM10 levels were associated with statistically significant reductions in lung function as measured by reductions in PEF and increased reported symptoms of respiratory disease and asthma medication use” (1).
While traditional regression approaches only tell us about the effect of a change in pollution on the child with the average initial PEF, we thought that it might be interesting to know the effect of pollution level changes on the distribution of children. If an increase in PM10 levels decreases PEF more drastically for the children with the initially lowest PEF levels then the seriousness of pollution as related to respiratory disease in children may be understated. In order to see how the pollution levels affect children with different initial PEF measures we used the data from Dr. Pope’s paper and a similar model, controlling for the same factors, and performed a quantile regression.
We ran several different regressions with different measures of pollution, as Dr. Pope had done, and decided to use the Orem measure. We also tested the sensitivity of the results to the number of iterations used when bootstrapping the standard error and found that the results did not differ too much.
So to answer our initial question, do the 10% of children with the lowest initial PEF measures have a larger percentage change in PEF measure than those with the average initial PEF measures? We found that they do indeed. While an increase in the PM10 levels by 10 units causes approximately a 6% decrease in PEF among children in the .5 quantile (the children with the average initial PEF) it is associated with a statistically significant decrease of 12.68% in the .1 quantile (the 10% of children with the lowest initial PEF). While not as extremely different from the average effect, the 8.5% decrease for the children at the .25 quantile is also statistically significant and larger than that at the effect at the mean (Table 1).
In looking beyond the original study, we have found that that an increase in the PM10 levels is not only with a decrease in PEF measures for children on average, but with a larger percentage change in PEF for the children with the lowest initial measure than the average.