According to standard conventional political wisdom, the way to win an election is to motivate one's base voters while simultaneously pursuing swing voters. However, in this year's GOP nomination race storylines have focused very heavily on the first part of that equation and very little on the second. We have heard again and again that the contest centers around a moderates vs. conservatives divide in the Republican party. More specifically, we have heard that John McCain has been more successful in the primaries and caucuses thus far because McCain has been able to unify party moderates behind his campaign to a greater degree than Mitt Romney has been able to unite party conservatives.
There may be some truth in the "moderates vs. conservatives" theme. If, for example, Mike Huckabee had dropped out of the race before the Florida primary his votes may have gone disproportionately to Romney (though that is far from certain, at the least) . However, the theme is also somewhat misleading because one of its fundamental assumptions -- that the GOP is divided into moderate and conservative voter blocks -- appears wrong, or at least incomplete.
In conducting exit polls from the five strongly contested states so far -- Iowa, New Hampshire, Michigan, South Carolina, and Florida -- pollsters did not simply ask GOP voters whether they were "moderate" or "conservative" in their views. Instead, they used a more informative five-choice model, asking voters whether they were "very liberal" (not too many responded that way, as you might expect), "somewhat liberal", "moderate", "somewhat conservative" or "very conservative."
Perusing the exit poll figures from the major contests, I put together a table listing the percentage of voters who labeled themselves "moderate," "somewhat conservative," or "very conservative" in each race.
As you can see, in each race the "somewhat conservative" voters subgroup was either the largest group or a close second.
More interesting lessons come from how those somewhat conservative voters voted (or, at least, how they told the exit pollsters they voted). The following is a table showing the candidate who won "somewhat conservative" voters in each state and that candidate's margin of victory over his next closest competitor for those voters, along with the data for the overall winner of the state (from actual vote counts).
|Iowa||Huckabee +7||Huckabee +9|
|N.H.||McCain +3||McCain +5|
|Michigan||Romney +3||Romney +9|
|S.C.||McCain +2||McCain +3|
|Florida||McCain +3||McCain +5|
Obviously, the data for the winners and margins of victory in the "somewhat conservative" voter block fits fairly neatly with the overall results of the actual voting. But what about "moderate" and "very conservative" voters? Here are the results among all three subgroups, compared with the overall actual vote results.
|Moderate||Somewhat Conservative||Very |
|Iowa||tie (Mc/Rom)||Huck. +7||Huck. +12||Huck. |
|N.H.||McCain +17||McCain +3||Romney +24||McCain |
|Mich.||McCain +6||Romney +3||Romney +24||Romney |
|S.C.||McCain +30||McCain +2||Huck. +19||McCain |
|Fl.||McCain +22||McCain +3||Romney +23||McCain |
While the winner of each state won the "somewhat conservative "voter subgroup everytime, the same is obviously not true for the winners of the "moderate" and "very conservative" blocks . Moreover, in each state the winning margin among "somewhat conservative" voters was closer (usually much closer) to the winning margin among all voters than either of the other subgroups .
What does all this suggest?
That "somewhat conservative" voters (for lack of more narrow targeting), far from united for one candidate or another, are essentially a swing voter group in the race.
That point, if it's true, rather throws a monkey wrench into many assessments of the race between Romney and McCain. Many pro-Romney commentators (or, put more accurately, anti-McCain commentators) have been asserting that Romney is the leading conservative candidate in the race. The data shows that to be true among "very conservative" voters, but things are far more competitive among "somewhat conservative" voters. Indeed, while Romney might or might not be the frontrunner today if Huckabee had not competed for "very conservative" voters in New Hampshire or Florida, he would almost certainly be the frontrunner had he been able to win "somewhat conservative" voters in those states.
So why has Romney done so much more poorly among "somewhat conservative" voters than among "very conservative" voters? My educated guess (or perhaps not-so-educated guess): "somewhat conservative" voters tend to feel much less strongly about McCain's alleged heresies on illegal immigration, or even find themselves in relative agreement with McCain, than "very conservative" voters. (Full disclosure: I'm one of those voters who agrees with much of McCain's current position). Moreover, in my view many conservatives are unconvinced that Romney himself has a core of conservative principles.
Whatever the reasons for Romney's underperformance among "somewhat conservative" voters, he probably can't win the GOP nomination if it continues.
(As I publish updates on various results tonight, I'll also point to the exit poll data on what "somewhat conservative" voters did.)
Update: Did the best I could to fix the presentation of the tables.