"These findings build on and strengthen the conclusions of Donohue's earlier research, which only used data through 2006," Ingraham writes. "In addition to having nearly two decades' worth of additional data to work with, Donohue's findings also improve upon (the 1997 study) by using a variety of different statistical models, as well as controlling for a number of confounding factors, like the crack epidemic of the early 1990s."
"These new findings are strong," Ingraham writes. "But there's rarely such a thing as a slam-dunk in social science research. Donohue notes that 'different statistical models can yield different estimated effects, and our ability to ascertain the best model is imperfect.' Teasing out cause from effect in social science research is often a fraught proposition." (Read more)