It's hard for me to decide what to make of our polling in last night's races in Colorado, Minnesota, and Missouri.
On one hand our numbers didn't come very close to giving a precise measure of the final outcome. Rick Santorum did much better in each state than our numbers had projected.
On the other hand we were the only polling company that did any surveys in any of those three races. No one else even tried because of the difficulty in identifying the electorate for such low turnout caucuses and beauty contests. And although they didn't measure the full magnitude of it, our surveys did pick up that there was some major momentum for Rick Santorum in these states. There weren't really any other indicators that was coming- Santorum had done poorly in every contest since Iowa- and if not for our polls last night's results really would have come as a shock to people.
One is that he was getting stronger every day. When we added Monday's interviews to Saturday interviews we had conducted he gained 7 points on Mitt Romney in Minnesota and 4 points on Romney in Colorado. If we had only released our Monday interviews in those states our numbers would have been closer, and he clearly continued to have momentum on into Tuesday. We prefer multi day field periods, but this was a situation where things were moving fast and a one day poll actually would have come closer to the outcome.
The other thing I think may have had an impact is how volatile the electorate in these states was. 38% of voters in Missouri, 35% in Minnesota, and 31% in Colorado said they were still open to changing their minds about who to vote for. That compares to only 21% in Nevada, 19% in Florida, 22% in South Carolina, 24% in New Hampshire, and 24% in Iowa on our final polls in those states. A lot of folks who told us they were voting for Gingrich/Romney/Paul clearly either didn't show up or changed to Santorum in the closing stretch. Gingrich in particular did far worse than our polling had predicted and it's possible his folks jumped ship when they saw Santorum had a much better chance at defeating Romney.
Take aways from this? On the whole I think we did some public service by polling these states. It kept the Santorum surge from being a complete and total shock. But there's a reason none of our peers tried to poll these races- they may just be impossible to survey with much precision. And because of that we're probably done trying to poll caucuses and will stick to bigger state primaries for the duration of the Republican contest.