Mississippi exit polls wrong on Mitt Romney, but still reveal much about voters

» View our political blog, with live updates and analysis of the GOP presidential nomination process.

Long before the polls closed Tuesday afternoon, the Drudge Report teased readers with hints that Mitt Romney had won Mississippi. When the exit polls were released at 8 p.m. Central, it looked like an extremely tight race.

The exit polls had been strikingly accurate for the past two months, and here the Mississippi exit poll showed the two leaders within a percentage point of each other. The largest error in the past six contests had been half a percent.

But as the votes were counted, Rick Santorum jumped out to a lead. Observers who remembered Ohio figured it would dwindle. Ohio had brought in its urban Romney-friendly precincts last.

But Santorum did not fade. The lead held, and Romney came in third, three percentage points behind Santorum.

Two percentage points may not seem like a huge error, but in the narrow game of perception, it makes a difference. The error pushed Romney from a near tie for first into a narrow third-place finish.

To be fair, Mississippi is a state Romney should not have seriously contested, based on demographics. It was, as he put it, "an away game." In fact, the 30 percent he earned in Mississippi exceeded his take in the last four Bible Belt states this month, where he averaged 28 percent.

But close observers still ought to have suspected something was amiss as soon as they peeked inside the exit polls.

As we noted last week, the most potent question in exit polls this year is whether it matters to you that the candidate "shares your religious beliefs." Those who say it matters a "great deal" tilt heavily against Romney.

The proportion of the voting population heavily oriented by religious belief varies by state. In Arizona, only 20 percent were oriented that way, while in Alabama it was 45 percent. The four Bible Belt states that voted this month averaged 39 percent.

Mississippi actually had the largest group of "great deal" religious voters of all the recent states at 47 percent. But shockingly the exit polls Tuesday night said that Romney had claimed 26 percent of those voters — 11 points above his Bible Belt average.

In the last previous six races with exit polls (Ariz., Mich., Okla., Georgia, Tenn., Ala.) the percentage of the total vote that both voted for Romney and said religion mattered a "great deal" averaged 5.8 percent. The low was 3.3 percent in Georgia, the high 7.2 in Alabama.

In Tuesday night's Mississippi exit poll, that number was 12.2 percent — more than double the recent average. Something was wrong here.

The trouble, of course, is that the explanation proves a little too much. We only need to account for a missing 2.5 percent, and instead we have found over 6 percent of potential error.

So perhaps Romney outperformed his earlier score with this demographic as well. The fact that he did a little better in the demographic in Alabama suggests he might have gained some headway.

In any case, it remains intriguing that the best leverage for explaining just about anything this primary season — in this case the exit poll error — remains the "great deal" religion question.

Eric Schulzke writes on national politics for the Deseret News. He can be contacted at eschulzke@desnews.com.