This political scientist estimated politicians' beliefs via 100 million campaign donations

Would Scott Walker be the most conservative major-party presidential nominee in decades? That's what a pair of charts rocketing around the political science blogosphere purports to show.

But here's the interesting part: the charts aren't tracking what the candidates have done — how they've voted or what they've said. Instead, they're based entirely on who's given them money.

The metric aggregates many donors' beliefs about candidates

It might not be intuitively clear why looking at donors would tell us anything useful about what candidates actually believe. But Adam Bonica, the Stanford political scientist and cofounder of the site Crowdpac who created the metric used in the charts, argues that fundraising data allows for a better quantitative comparison among presidential candidates than past votes or positions do.

That's because there are over 100 million records of political donations over several decades — donations from people who, presumably, have their own views on issues and donate mostly to candidates who share those views. "It's an indirect measure based off of aggregating a lot of people's beliefs about a candidate," Bonica told me.

But it's important not to draw the wrong lessons from this metric. Winning the enthusiastic support of right-leaning donors doesn't necessarily imply that a candidate can't appeal to swing voters, too. So here's a guide to how Bonica's metric works — and what it could mean not only for Walker's 2016 hopes, but the emerging presidential contest more broadly.

Following the money

A donor is very unlikely to give to both ex-Rep. Michele Bachmann (R-MN) and Sen. Bernie Sanders (I-VT). (Tom Williams and Bill Clark, CQ-Roll Call Group / Getty)

The assumption behind Bonica's metric is that political donors have certain beliefs, and they'll overall tend to give to candidates who share those beliefs. Certain donors will mainly favor candidates on the far left or right, others might give broadly to all Republicans or Democrats, and others could give to mainstream candidates from both parties.

The point is, overall, donors with similar preferences tend to concentrate their giving among similar candidates. "If a donor gives to two candidates, those candidates will on average be more similar in their policy views than they would be at random," Bonica says. "It's likely that you'll see someone donate to both Bernie Sanders and Elizabeth Warren. But it's extremely unlikely that you'll see anyone give to both Bernie Sanders and Michele Bachmann."

With that in mind, Bonica crunched the numbers from political donations in federal and state races going back to 1979. He used a statistical method to plot both donors and candidates on a left-to-right axis, based on how various donors and donations clustered together around certain candidates. The end result, he says, "is this ideological spectrum, essentially, that corresponds pretty well to other measures that have been developed."

The inspiration: DW-NOMINATE

The 2013-14 DW-NOMINATE scores ranked Mike Lee, Rand Paul, and Ted Cruz as the first-, second-, and fourth-most conservative senators. (Chip Somodevilla / Getty)

Bonica was inspired by the work of his advisers Keith Poole and Howard Rosenthal, who developed a similar and commonly used political science metric measuring the ideology of members of Congress, from left to right — the "DW-NOMINATE" score. Their groundbreaking research measures partisan polarization and ranks members of Congress from most liberal to most conservative.

The strength of Poole and Rosenthal's approach is that no subjective decisions have to be made about which particular votes signify liberalism and which conservatism. DW-NOMINATE instead looks at how often various members of Congress vote with members of the opposite party and how often they dissent from their own party's preferences.

Though not all issues fit neatly on a left-right spectrum — some proposals may be supported by the far-left and far-right, or just by moderates in both parties — the vast majority of those that come to a vote in Congress do. So if a Republican frequently votes with Democrats and against her fellow party members, she'll be scored as a centrist. But if a Republican often dissents from his party line and isn't joined by Democrats, he'll be scored as very conservative. Here's the ranking from the 2013–2014 Senate, which lists Mike Lee (R-UT) and Rand Paul (R-KY) as the most conservative senators and Susan Collins (R-ME) and Lisa Murkowski (R-AK) as the most moderate Republicans.

The problem with applying that metric to the presidential race, though, is that the candidates aren't taking the same set of votes. Particularly, governors are signing or vetoing bills that are usually unique to their states and have already gotten through the legislature, rather than voting on anything. The underlying data is so different that a quantitative comparison doesn't make much sense.

Bonica's modification: the CFscore

Campaign finance data, though, isn't limited in the same way — because most donors to state races have also given to national campaigns at some point. "In any given state," Bonica writes, "between 70 to 90 percent of contributors who fund state campaigns also give to federal campaigns, providing an abundance of bridge observations."

So, Bonica essentially used statistical techniques similar to Poole and Rosenthal's, but applied them to campaign finance data, instead, to come up with what he calls "CFscores."

it's an incredibly rich and complex data set

He also used this data to compare the ideologies of candidates from different years and even decades. "Because you both have candidates and donors that are active in multiple cycles, it provides this sorta glue from one cycle to the next," he says.

When one thinks of political donors and donations, huge sums from millionaires and billionaires often come to mind. But the way Bonica's model works, donations from many small donors can be crucial in determining a politician's CFscore.

"The way the model incorporates the data is if you get $1 million from one person and $10 from 500 other people, the 500 people who gave you $10 would be far, far more important in determining your score," Bonica says. That's because each of those individual donors likely has a unique donation history that can give information about which other candidates hold similar views.

So the more than 100 million political donations that go into Bonica's model make for an incredibly rich and complex data set — one containing a great deal of information on who gives to whom, with many implications for what that might mean.

One drawback: politicians who get makeovers

Mitt Romney (Jim Watson/AFP/Getty)

According to Bonica, most politicians' CFscores are pretty stable throughout their careers. But there's one big recent exception — Mitt Romney, whose score changed quite a lot as he ran for different offices.

"Romney's score was much more moderate when he ran for Senate and then governor of Massachusetts," Bonica says. "Then it swung as he moved to the center of the party in 2008, and it was more stable after that."

Changes like this in Bonica's data occur most often when a politician moves from state elections to a presidential bid. "There's this chance for presidential candidates to redefine themselves as they become a national candidate," Bonica says. They can change their positions or their emphases and start raising money from different people as a result. If that happens, the change will be evident in the data — but not until new fundraising information comes in. "So there's some forecasting in trying to apply this to 2016 now," he says.

What it means for 2016

With all that in mind, let's take a look at what the data does show for the presidential race, remembering that it reflects these potential candidates' past campaigns because current fundraising figures haven't been released yet. Many findings are expected — Jeb Bush is somewhere between John McCain and Mitt Romney, which sounds accurate. Candidates like Mike Huckabee, Ted Cruz, and Rand Paul fundraise from the right edge of the party, while Chris Christie does so from the left. (We'll see if Christie manages to pull off a Romney-like makeover if he runs nationally.)

One unexpected result is that Rick Santorum — generally thought to be a far-right conservative — has donors more toward the center of the party, on average, than candidates like Bobby Jindal, Mike Pence, and Marco Rubio (as well as Walker). This could reflect Santorum's many years of House and Senate fundraising, when he was a leader in the party, raising from its mainstream donors rather than being pigeonholed as a social conservative candidate, as he is now.

The most attention-getting finding, though, is that Scott Walker is rated as extremely conservative. Part of what the data is picking up on here is Walker's wide appeal among many conservative small donors, as well as large ones. "He's raised a lot of money from a lot of people. I think in 2012 he had 125,000 donors to his campaign, which is almost unheard of for a gubernatorial candidate," says Bonica. And these donors "seem to be drawn from the right wing of the party. Typically, people who draw from a lot of small donors tend to be more at the extreme."

And here's another look at the chart up top from Jason McDaniel, a professor at San Francisco State, which shows CFscores for every open seat GOP presidential winner going back to Reagan, as well as Jeb Bush and Scott Walker. The red dots show the average scores for presidential candidates that cycle, with Reagan as the only conservative donor candidate to defeat rivals backed by more moderate donors, on average.

So Jeb's donors look a whole lot like his dad's, Bob Dole's, McCain's, and Romney's, in terms of both overall score and placement relative to that year's field. Walker, on the other hand, has the most conservative group of donors overall and is about as far to the right of his rivals' average score as Reagan was.

McDaniel's overall takeaway from this chart is that, "If the recent history of Republican nomination contests is any guide, the party is likely to decide that Scott Walker is too ideologically extreme to be the Republican nominee in 2016." Walker has a great counterargument to this, though, in his electoral history. His supposed extremism didn't prevent him from winning gubernatorial elections in a blue state three times by more than 5 percentage points.

Beyond that, it's not evident to me that Walker's overall positions are far more extreme than, say, Jeb Bush's. Rather, what the CFscore shows is that Walker's shown a capability of winning the enthusiasm of very conservative donors, including many small donors, in a way Jeb hasn't.

Conservative small donors, of course, vote in GOP primaries. So the fact that Walker's won many of them over without tarnishing his general election prospects so far makes him a very appealing primary candidate. That is, his score here might not reveal his weakness — but instead, his strength.

Explanatory journalism is a public good

At Vox, we believe that everyone deserves access to information that helps them understand and shape the world they live in. That's why we keep our work free. Support our mission and help keep Vox free for all by making a financial contribution to Vox today.

One-Time Monthly Annual

$95/year

$120/year

$250/year

$350/year

Other

$

Yes, I'll give $120/year

Yes, I'll give $120/year

We accept credit card, Apple Pay, and Google Pay. You can also contribute via