I think this is a very bad paper and these graphs are a perfect example of why. These graphs give the definitive impression that the counties sans AT&T coverage and the counties with it had the SAME birth-rate trend pre-iPhone and then diverged after it.
