Hugh Hewitt published a very thoughtful piece recently breaking the Republican Party primary electorate into “four types” of voters. Since this sounds similar to my “Four Faces” breakdown, a number of people have asked me if we agree. We do not.

Mr. Hewitt’s breakdown is based on his subjective assessment of the motivations different groups of GOP voters have when selecting their preferred candidate. Thus, he breaks down the electorate into “True Believers,” “Buckley Voters,” “Center-Right Governing Conservatives,” and “Uniques”.

The first group likes red-meat conservatives who talk about principles – Hewitt lists Cruz, Huckabee, and Santorum (along with Carson and Trump) as candidates this group backs. The second group looks for the most electable conservative, following the late Bill Buckley’s dictum that this is the type of person conservatives should support. Hewitt says Perry and Walker voters came from this group, and they are now likely to flow to Fiorina or Rubio.

The third group “think themselves more realistic about what can and cannot be done by government” and favor people with “less combative” and “more inclusive” rhetoric. Bush, Kasich, and Christie are the candidates favored by this group.

The final group are the people who favor their person for candidate-specific reasons. Paul and Graham are this cycle’s unique candidates for Hewitt.

There’s a lot of perceptiveness in this, as Hewitt captures something not often appreciated by most members of the media: the general preference for temperamentally calm conservatives among a majority of GOP voters. It’s a big reason why the eventual nominee is often considered to be “boring” by movement conservatives: they want fire and flash, but the average Republican wants a calm, reassuring figure at the helm.

I disagree with Mr. Hewitt, however, in thinking that disposition is the primary fissure between GOP voting blocs. My forthcoming book and the article upon which it is based examine four cycles worth of primary exit polls and found that ideology (mediated by religiosity) was the most important element in determining how the GOP contest broke down.

My four groups are “Very Conservative Evangelicals,” Very Conservative Fiscal Cons,” “Somewhat Conservatives,” and “Moderates”. The first group likes candidates who place religion and social issues as a primary, if not the primary, value. The second group likes people who want to dramatically cut taxes and spending. The third likes people with governing experience, conservative views, and a calm disposition. The fourth likes non-conservatives who speak their mind and do not place a high value on religion in the public square.

Based on the exit poll data, this approach quantifies how many people are in each group, both nationally and in each state. That allows one to forecast which type of person will win the early states, which determines who gets “winnowed in” to the middle stages of the race. This also one to predict who will emerge as the winner in the final one-on-one showdown, as there are clear patterns as to where voters in each ideological faction migrate when their favorite drops out.

Historically, this method allows one to understand why the “movement” conservative candidate has usually been a social conservative. In the past, that faction was both larger than the fiscally-focused very conservative faction and were especially powerful in the Iowa caucus. That regional strength let that faction’s favorite win or exceed expectations in every caucus since 1996, giving the social conservative a huge early media boost.

This approach also accounts for the continued influence of non-conservatives in the GOP. Moderates are about 25–30 percent of the national electorate and dominate New Hampshire. Dole, McCain, and Romney won their respective nominations by winning large margins among moderates and winning pluralities among the sober-minded somewhat conservatives. The only exception to this rule — George W. Bush — won with united movement support but was himself sober-minded enough to win majorities of the somewhat conservatives in a one-on-one showdown with then-maverick John McCain.

Understanding the role of moderates in the 2016 race allows one to see why John Kasich is emerging as a potential breakout candidate. He is consciously trying to recreate the moderate–somewhat conservative alliance that worked for Dole, McCain (2008) and Romney (2012). That’s why he is focusing on New Hampshire, the early voting state with the largest share of moderates. And the polls to date show his strategy is working. Three of the four most recent New Hampshire polls with ideological breakdowns (here, here, and here) show Kasich’s support is heavily titled toward self-described moderates; he polls between six and nine points higher than his statewide total among those voters. He is, predictably, least popular among self-described “very conservative” voters, getting 2 and 1 percent of them in the two polls that break conservatives into those subgroups.

Perhaps this is the cycle where the old patterns fade away and a new model of GOP intra-party dispute asserts itself. If so, I think Mr. Hewitt’s approach is likely to prove accurate. But for now, I’m sticking with my data-based model.