As discussed above, we believe that a major contribution of our study is that we can improve the accuracy of our pollution measures because we have the mother’s exact addresses.
In Table 7 we offer two investigations of this claim.
If being closer to a monitor improves measurement, then being farther from a monitor should yield weaker results.
Table 7 shows that this is indeed the case: we do not find significant effects on health at birth (or, not shown, on infant mortality) for mothers 10–20 km from a monitor.
Similarly,studies often do not have an exact address of the mother but only the zip code of residence, and therefore assign pollution to the zip code centroid using an inverse distance weighted average of monitors near the zip code.
In the last three columns of Table 7, we assign pollution to the mother assuming we only know her zip code.
In this less precisely merged sample we find generally smaller estimates that are statistically insignificant.
Both of these results are consistent with improved measurement from knowing the mother’s exact address.