Actually, we had to do more than that. Since the nomination process is sequential — states vote one at a time rather than all at once — we had to determine, empirically, how much the results of one state can affect the rest.

The answer in the case of Iowa is that it matters a lot. Despite its demographic non-representativeness, and the quirks of the caucuses process, the amount of media coverage the state gets makes it far more valuable a prize than you'd assume from the fact that it only accounts for 41 of the Democrats' 3,979 pledged delegates.

More specifically, we estimate — based on testing how much the results in various states have historically changed the candidates' position in national polls — that Iowa was the second most-important date on the calendar this year, trailing only Super Tuesday. It was worth the equivalent of almost 800 delegates, about 20 times its actual number.

Which states will produce the biggest bounces?