I’m thinking that any votes lost by Santorum won’t be going to Romney.
Well, at his peak on February 19th, Gallup tracking had Santorum at 36, Romney at 26, and Newt at 13.
Today’s release (Feb 28 final date) has Romney at 33 (+7), Santorum at 25 (-11), Gingrich at 16 (+3).
So it appears that Romney gets over twice as many “lost Santorum votes” as Gingrich gets.
Which meshes with what we see in most preference polls, and mirrors what we see in states where Gingrich came in 2nd, vs states where Santorum comes in 2nd. Namely, if Santorum is in 2nd, the combined Santorum/Gingrich vote is better than when Gingrich comes in 2nd. In fact, there are NO cases yet where Gingrich beat Santorum but came in 2nd and the combined Gingrich/Santorum vote beats Romney. Whereas in Michigan, even though GIngrich only got 6%, if you combine him with Santorum they beat Romney easily. (Arizona the two of them lost to Romney, but kept him below 50%: in Nevada, when Gingrich came in 2nd, Romney was over 50% total).
One explanation for that is that, in fact, Santorum votes tend to split evenly or more toward Romney, so if Santorum is pushed down, Romney goes up by more than Gingrich, and (as in Florida, Nevada, and New Hampshire) the total Newt/Rick vote loses to Romney.
But if Santorum is in 2nd, as in Michigan, the combined vote of Rick/Newt is higher, because Newt’s voters tend to move to Santorum.
I don’t know why. I just know that it seems to be holding true, and that this makes it harder for Gingrich to win, because pushing Santorum out tends to help Romney at least as much as Newt.
Maybe that will change going forward.