Counterfacturals are all the rage in economics, and I guess we could make probablistic inferences about the likelyhood of certain outcomes in politics, too, but the notion that we can write history backwards is just worth the time we spend on it.
We have a bias in favor of causality. If we associate an event with a cause -- even if the association is later proven to be tenuous or non-existent -- we'll have a hard time breaking effect from cause in our minds. Counterfactuals are intuitive in that respect. With an existing association, we can simply add a "cause" to our logic chain and then adjust according.
IF John Edwards had been caught earlier and dropped out before the caucuses, THEN Hillary Clinton would have won Iowa BECAUSE she and Edwards shared the alliegiance of white working class voters, union voters, etc. Well first, even as a basic syllogism, this doesn't really work because the evidence isn't there.
Second, even if we could establish some sort of reverse causality, there's still an infinite number of possible pathways that could lead in an infinite number of directions. If Edwards's affair had been revealed earlier, such as, say, right after it began, Mark Warner might have stayed in the presidential race. Indeed, if Mark Warner had decided to run for president, then he might have filled the anti-establishment void that Barack Obama jumped into. If, If, if.
| 
