MICHIGAN — Election night demands instant coverage from reporters—not just to name a winner, but to define how voting patterns shape the victory. And often, that process of definition begins before the results are even in. Here in Michigan, well before the votes were tallied in Tuesday’s Republican presidential primary, journalists had begun to report on the contest through the frame of class division: the filthy-rich Mitt Romney was expected to have the support of wealthy voters, while comparatively blue-collar Rick Santorum would find support among working-class voters.

A pre-vote piece in The New York Times, for example, suggested that Santorum’s tactic of “aggressively pushing to win over blue-collar workers here” could “make a critical difference,” while an Associated Press report said Romney struggled for white working class support. The Los Angeles Times pushed the angle hardest, declaring the Michigan race an “all-out class war” in language that bordered on the mythic: Romney was “heir to a storied Michigan political dynasty” while Santorum was “waging a determined challenge” and “flaunting the fieriest populism” seen in years. Oh my!

And as Romney edged out a victory, exit polls indicated there was some truth to the class divide story: Mark Blumenthal, senior polling editor of The Huffington Post, provided an election-night write-up detailing how Romney bested Santorum by 26 percentage points among Republicans making more than $200,000 per year, while trailing Santorum among those making less than $100,000. Class division became a prominent angle in national coverage: Forbes gave Santorum credit for aiming “below the elite” in a close race, and The New York Times declared that Santorum “successfully played up his own working-class credentials” in Michigan.

This is a sensible, newsworthy story that appears supported by the exit poll data—but it’s not the only possible story, and not necessarily the most interesting one. Another significant angle is the ideological split, which Blumenthal explored in a smart and thorough follow-up on Wednesday:

In Mitt Romney’s narrow victory in Michigan’s Republican primary on Tuesday, he received a slightly greater share of the vote (41 percent) than he won four years earlier (39 percent), but that slight increase obscures a larger, ongoing story. Since 2008, Romney has consistently gained with moderate Republicans, but has typically lost support from the most conservative voters in the Republican base.

Four years ago, former Massachusetts Gov. Romney positioned himself as the conservative alternative to Sen. John McCain (R-Ariz.). As such, according to the National Election Pool network exit poll in Michigan, Romney defeated McCain by a wide margin (48 to 11 percent) among Michigan Republicans who described themselves as very conservative. McCain prevailed narrowly (38 to 34 percent) among Republicans in the state that identified as moderate or liberal.

On Tuesday, the support for Romney realigned. According to the 2012 NEP exit poll in Michigan, Romney lost very conservative primary voters to former Sen. Rick Santorum (R-Pa.) by 14 percentage points (36 to 50 percent), but carried both moderate and liberal voters (39 to 33 percent) and those who identified as somewhat conservative (30 to 32 percent).

Using data tables to back up his point, Blumenthal points out that the ideological shift is not unique to Michigan. Rather, the “very conservative” subgroup has been Romney’s worst in all of the early contests for which exit polling is available—even though in 2008, when Romney was presented as the “conservative alternative,” the reverse was true.

While this shift has not been a major theme of campaign coverage, Romney’s status as this cycle’s “moderate” has been. So why did ideology recede in favor of the class gap in post-Michigan coverage?

There are several possible reasons. One is that, as Blumenthal notes, the ideology story is consistent. On the other hand, the class gap is new, and it may carry over to demographically similar Ohio. The class story also taps in to perceptions about the candidates, which allows reporters to frame the narrative in personal terms. And it was primed by the pre-vote coverage: reporters may have been looking for it, because it was a story they’d been writing and reading about in advance of the vote.

Blumenthal offered another explanation: most reporters don’t have access to raw exit poll data. The television networks that fund the polls do, and some major newspapers subscribe to “a la carte” data on election night, but, he said, “the rest of us” do quick-hit analyses from partial data gleaned from polling consortiums and other news reports. “It’s very limited,” he said.

And with that limited access and limited time, the class divide story was easier to identify. Blumenthal had the same data access for his examination of shifting ideologies, but it required more hours of sifting than election night coverage allows. His ideology story was published late the next day.

Again, this isn’t to say that the “class divide” story is wrong (even if some of the coverage that adopted that frame was overwrought). But the risk is that one storyline takes on outsized importance—and thus shapes future coverage—after hooking into election-night headlines for arbitrary reasons.

So, for the political reporter with limited access to raw data and the screaming deadlines of the 24-hour news cycle: What to do?

Blumenthal offers one “long-haul” suggestion: journalists should read the work of political scientists who break down voter patterns—people like Simon Jackman, who recently examined the role of social conservatives in the Republican primary, or Jonathan Bernstein, whose blog often focuses on campaigns and elections. The insights of these observers, said Blumenthal, should “guide our coverage … they have the time, skill, and access to data to do the analyses that we don’t. And they’re not limited by the tight deadlines we have in a 24/7 news cycle.”

More immediately, news organizations can plan for the follow-up, as a way to guard against getting trapped in the narratives they’ve constructed. While Blumenthal was among many who picked up the class angle on election night, he went back to the numbers to tease out the ideology story, even as other reporters moved on to advance coverage of Super Tuesday.

And what he found suggests opportunities for a further follow-up, one that doesn’t emerge from the “class divide” frame. Blumenthal shows Romney’s support comes from the more moderate part of the GOP base, a reversal from 2008. But has Romney’s political platform become proportionally more moderate? Or—as seems more likely—is Romney losing support from conservatives even as he runs on a platform that is further right? Why is the ideological shift among voters moving in a different direction than Romney’s shift of ideology? The answer might tell us something about how candidates are presented and perceived—and also about the shape of the choice before voters this fall. I’m sure I’m not the only one who wants to know.

If you'd like to get email from CJR writers and editors, add your email address to our newsletter roll and we'll be in touch.

Anna Clark is CJR's correspondent for Michigan, Wisconsin, Ohio, and Pennsylvania. A 2011 Fulbright fellow, Clark has written for The New York Times, The American Prospect, and Grantland. She can be found online at www.annaclark.net and on Twitter @annaleighclark. She lives in Detroit.