I have been pursuing the issue of electability and trying to explain to readers why small differences can have big consequences. In
this post I give an example in which there are 17 contested states and the candidate must win 9 or more to win the general election. If the probability of winning a contested state is increased from 50% to 60% the probability of winning the general election increases from 50% to 80%.
There are other ways in which electability can influence probability of winning. More electable candidates will reduce the number of states in the "toss-up" catagory that are required for victory by changing a toss-up into a solid win column or by making a state that had been a solid win for the opponent into a toss-up.
As a baseline consider the situation in which a candidate has a 50% probability of winning a toss-up state, there are 17 toss-up states and 9 or more must be won in order to win the general election. As above, the probability of winning the general election is 50%.
Now suppose we choose a slightly more electable candidate who can take just one of the toss-up states and turn it into a sure victory and take one of his opponent's sure states and turn it into a toss-up. For this candidate there are 17 toss-up states and only 8 or more are required to guarantee winning the general election. The probability this new candidate wins the general election jumps from 50% to 68%.
As before, small changes in electability are especially magnified in close elections in which many different states are in play.
Bottomline: those who would claim that candidate electability should NOT be an issue for primary voters must believe either (1) the general election is not going to be close because any Democrat is sure to lose; or (2) the general election is not going to be close because any Democrat is sure to win; or (3) the general election will be close but only in a very few states.