Princeton Election Consortium

A first draft of electoral history. Since 2004

The Real Problem With That Chart

April 13th, 2015, 11:03pm by Sam Wang



In today’s dustup between FiveThirtyEight and Vox, the press is missing the point. This is often the case, since most reporters understand mud-throwing better than they understand data.

I am not interested in squabbles over whether it is kosher to show someone else’s graphic. That ship sailed ten years ago when blogging got big. See Ezra Klein today; it’s what aggregators and commentators do. The real story is that the original interpretation is quite possibly wrong. Go read what Matt Yglesias actually wrote!

The bottom line, in two sentences: 1) Hillary Clinton has Presidential-level name recognition, which nearly the entire GOP field would kill for. 2) Jeb Bush is starting off as damaged goods, but most other Republicans are not.

Follow me down to understand why.

First, let me re-plot the data. I do not particularly want to get sucked into the vortex of mud-throwing!

Note that I have not done the cute 45-degree rotation that was the signature of the original FiveThirtyEight graphic. That was clever…maybe too clever, because it obscured important features in the actual data. When the analysis is the story, that can be a danger sign.

Here are three major features I see in the data set.

  1. Hillary Clinton has massive name recognition. She is as well-known as a sitting President.
  2. As of today, Hillary Clinton’s favorability is 13 to 22 points higher than every Republican in the race.
  3. The best-known Republican, Jeb Bush, matches Hillary Clinton’s unfavorability, but lags her in favorability by 15 points. To match her net favorable-minus-unfavorable number, he would have to win over people who don’t have an opinion by a ratio of 1.7 to 1. That is a huge challenge.

(There’s a fourth feature: why is Elizabeth Warren even on this chart at all? Did you guys get the memo? She’s not running!)

I do not know how any analyst can go from these facts to saying that “nobody in this race is popular.” I wonder if FiveThirtyEight feels business pressure to avoid saying things that are particularly favorable to one side or the other. They wouldn’t want to alienate half their readers.

All that said, it is certainly possible for some of these GOP contenders to catch up, especially the ones on the lower left corner of the graph. Ben Carson, Scott Walker, and Marco Rubio are starting with a relatively blank slate.

In my view, a major story here is that Jeb Bush is starting off as damaged goods. Jeb has no scandal, yet he is as well-known – and as unpopular – as Governor Chris Christie of New Jersey, who is facing a fairly adverse Bridgegate situation. And as I have argued before, Christie is not going to get the nomination.

Now think about the fact that Bush is about to enter a long primary season. Sure, he could recover after the dust clears on the GOP nomination. If I recall correctly (update: I do), Mitt Romney recovered in 2012 after the primaries ended. But at a minimum, Bush has some work ahead of him.

A serious challenge for all the GOP candidates is that none of them has a favorability above the low 30s. It’s been said that as voters align with parties rather than individual candidates, candidate qualities matter much less than they used to. What if the GOP brand is toxic? Certainly more voters are aware of “Bush=GOP” than they are of Jeb Bush’s policies as governor of Florida. So it’s possible. In that case, then as any candidate becomes better-known, he will stay on a line that looks like this:

If that line holds up, it means that as people get to know a GOP candidate, they will break unfavorable-to-favorable by a more than 3:1 ratio. If so, then the eventual nominee will end up where Bush and Christie are today.

On the other hand, it could be that after GOP primary voters settle on a nominee, that nominee can break through this pattern. Whatever one thinks of the 2016 election, it is unlikely to be a blowout for the Democrats. So something has to change in the Republicans’ favor. To my eye, Marco Rubio has the greatest capacity to shift this dynamic. I don’t think Ted Cruz is the one to do the job.

Finally, a word on the Vox issue: Possibly, they could have avoided this problem simply by re-plotting the data, as I have. It took me 20 minutes to do so, and that’s only because I draw my plots by hand in MATLAB and Illustrator – basically I am an artisanal graph-maker. Come on, Yglesias, show a little craftsmanship! I am kidding – I love you guys. And hey, you can link to my work anytime.

Tags: 2016 Election · President

33 Comments so far ↓

  • Amitabh Lath

    There is a lot of confirmation bias going on with Clinton’s poll numbers vs. the GOP candidates. I do not think she is at all a sure thing. I think Jeb Bush has a high probability of winning the R primary and depending on voter turnout demographics, winning the general election.

    Note that predicting turnout demographic is the achilles heel of polling. The hope seems to be that the bazillion likely-voter models will somehow straddle the true value, but we have seen in the past how everyone makes a mistake in the same direction.

    The evangelical voters that sat out McCain and Romney may well turn out in droves for the guy who went all out for Terry Schiavo. If he stays sane on immigration, and picks Suzanna Martinez as VP then he’s competitive in CO, NV and wins NM. And maybe the hispanic vote nationwide doesn’t break D as much as ’08, ’12.

    A lot of Clinton boosting counts on women voters playing the role that african-american voters played for Obama. I am not convinced, mainly because I don’t see Clinton bringing first time voters to the polls like Obama did.

  • Joseph

    In the world of paying blogs, what keeps you alive are “hits” and “comments”. 538 is a paying blog, while P.E.C. is a “labor of love”, if you will. But both blogs have gotten a substantial number of hits and comments out of this somewhat flawed comparison. As remarked earlier, for a lot of the blogosphere, there’s a monetary advantage to making something a contest. I really think that whole idea of trying to make people think a race is close can end up having an impact on the end results.
    Thankfully, there are places like P.E.C. to keep the paying blogs at least somewhat under control!

  • Lee A. Arnold

    Sam, hi. I agree that the whole story will be decided on the GOP side.

    I think Jeb vs. Hillary will come down to “George W. Bush Administration” vs. “Bill Clinton Administration”, — and thus: “Iraq War + rich tax cuts + growing deficits + attempt to privatize Social Security + financial crash” vs. “Bill’s economic record”, which even many moderate Republicans came to admit was good (after Clinton was out of office and it was safe for them to admit, of course).

    Jeb speaks Spanish, and could put Rubio in the VP spot, which would be competitive. But Jeb’s got a whale of a bad track record to live down, merely by family relation.

    Hillary is also has the best foreign policy experience by a long shot, although that isn’t foremost in the minds of most voters.

    I guess that Walker or Kasich (if he steps in) would have the best chances in the general, although Walker’s running the old Reagan playbook without a modern update, and Kasich should consider that the intellectual degradation of arguing with all the fundamentalists in his own party, just to run against such an overwhelming name-brand (and someone whom he probably agrees with) in the general, may not be worth it.

    I think in the end it will come down to, “Send the girls to Washington! The boys have screwed it up enough!” I think there will be a huge surge among female voters, particularly young ones (and the Hillary camp clearly thinks so, to judge from her rollout video).

    If Rubio plays his cards right, though, he’ll be elected President in ’20 or ’24.

    • Andrew Dzeguze

      Just a technical comment – Constitutionally, Rubio or Bush would have to change their residence to run on the same ticket. Cheney pulled it off in 2000 by claiming his long time Wyoming residency – even though he’d been living mainly in Texas for a decade and change, he could at least claim it with pretty decent legal grounds. Rubio would be hard pressed to do say he has residency in any state but Florida, and while Jeb could probably do it no top ticket candidate wants a week of stories about “abandoning” the state they were governor of.

    • Lee A. Arnold

      I thought that was an electoral college restriction, but you’re right it would be a mess because Florida could be decisive, and the electors would have to vote for the Dem VP.

    • Matt McIrvin

      The rule is that the electors from a given state aren’t allowed to vote for both a presidential and VP candidate from that same state. For that reason, running mates with official residencies in the same state are generally not done.

      I suppose if both of them came from a state they were expecting to lose anyway, it wouldn’t be an issue, but Florida is generally considered a critical swing state, so it’d probably be a consideration.

  • Lee A. Arnold

    Hillary’s negatives have been vetted for 30 years and won’t change. I think her positives will go higher. I predict she will get all the Democratic votes + all the serious foreign policy votes from moderate Republicans + about 1/2 of all the Republican women who aren’t seriously anti-abortion. A landslide, possibly historic.

    • Sam Wang

      That seems on the optimistic side for the Democrats. But certainly, by now voters think she is a known quantity.

      At this point, a major unknown is whether someone can survive the GOP primaries and get his own positives to rise. My current sense is that Jeb Bush can’t do it, but I could be wrong there.

    • ottovbvs

      I think you may be right about at least the gender issue which I’ve rather dismissed previously. A totally anecdotal experience today. I visited my local family owned garden center which is staffed by three generations of women from the same family. I always joke with them and call it girl town. During today’s exchanges I mentioned Hillary and it was universally “Yeah Right” including senior grandma and ma who said we’re Republicans but this is something different. I was surprised but went on my way with three boxwoods to repair winter’s ravages.

    • Matt McIrvin

      A phenomenon I noticed back in 2008, and that still seems to be the case, is that if you look at Internet political discussion fora, Hillary Clinton seems to be much less popular than she really is. You see a lot of people on both the left and the right bashing her, and it’s easy to get the impression that she’s widely despised. But if you look at poll numbers it’s a different story.

      That might well be a gender thing: Internet political blather is lopsidedly male.

  • ottovbvs

    My first look at Mr Wang this presidential election season and as usual he didn’t disappoint. Bush is the most likely Republican nominee but he’s potentially a huge liability because of both the R label and the inevitable re-litigation of the Bush 2 presidency. Overall my view of 2016 right now is fairly simple given polarization and demographics. It’s the turnout. For the last two presidential elections it’s been 62% and 58% respectively. No reason now, other than perhaps Democratic complacency, to believe it will be wildly different in 2016 is there? The largest Republican vote ever was 62 million in 2004 and it’s been 1 to 2 million less in the last two elections so it looks like the Republican ceiling is around 62 million. This leads to certain conclusions.

  • Amitabh Lath

    This plot is what, All People? Not Registered or Likely voters? Is it at least weighted for demographics?

    Yes it is nice to have high popularity numbers, but having big leads in CA, MA, NY etc (which is what these numbers are probably due to) doesn’t mean jack in the electoral math. It’s going to come down to OH, PA, and FL as it always does.

  • Adrian

    Adjust this graph to match the likely 2016 electorate – after Obama jams through a terrible trade deal and the right makes ~$2 billion in dark money investments in sliming and voter suppression – and things will look a lot more competitive.

  • San Fran Sam

    Couple of comments….

    First, Clinton hasn’t been Koched yet. so I expect her unfavorables to increase.

    Second, what is missing is the no opinion share. Carson has a total of about 40% fav+unfav and about 60% no opinion. so you could draw isolines indicating no opinion.

    Third, as someone else commented we are observing partisan identification as the candidate is more well known. It would be interesting to see a graph for each party identification, Dem, Rep, Ind.

    • Ryan

      Wait wait wait, she hasn’t been Koch’d yet? We’re talking about Hillary Clinton here, the most well-known politician of the 21st century, target of Republican attacks for the last 20-odd years.

      I’m not saying her numbers can’t be moved, but it’s not like there’s a lot of people who don’t already have an opinion that was formed in the face of (or because of) years and years and millions of dollars spent trying to bring her down.

    • ottovbvs

      Are you serious? Republicans have spent over 20 years demonizing her. How does it get worse than suggesting she murdered Vince Foster on which nonsense the oped page of the WSJ seriously opined for years. And then more recently we’ve had 2 years of Benghazi! and the email dust up. None of it created any traction for the simple reason that when it comes to smearing Clinton, the law of diminishing returns cut in years ago. She is inoculated in a way that none of the Republicans are.

    • Matt McIrvin

      Hillary Clinton’s positives have gone down, though it’s possibly just that she was temporarily less of a partisan figure during her Secretary of State days, but now that she’s once more a presidential candidate and the de facto Democratic standard-bearer, her favorability is a proxy for “do you support Democrats or Republicans?” again.

  • Jack Waldron

    Interesting take. I think, it’s all a bit overblown. While the fight for the Primary win will be very damaging to the eventual candidate, you’re all three ignoring the Elephant in the room. The Country is still, last I checked, mostly half Dem/half GOP. The GOP voters hate the Dems more than any GOP candidate. The fight is over those in the middle, like me. None of the GOP candidates give me a warm and fuzzy. Also, I understand Nate’s ire. It’s rude (You know, bad manners. Manners used to matter.), not to mention plagiarism, to take someone’s work and present it as your own. A simple reference or link back of where it came from was the “right” thing to do.

    • Amitabh Lath

      The first line of that original Vox article by Matt Iglesias reads:

      Nate Silver and his team at FiveThirtyEight put together this great graphic…

      So he mentions Nate and his crew in the first sentence and still it is somehow plagiarism?

    • Matt McIrvin

      Well, there are two fights. One is over voters in the middle, where Democrats probably currently have an advantage. The other is partisan turnout, which is where Republicans do very, very well and Democrats are spottier, especially in non-presidential elections. To some extent the strategies for dealing with the two are contrary.

  • Richard Kettler

    The problem with all of these charts is that they model increasing fame as moving along a line on which
    Favorability Rating = Unfavorability Rating.
    That isn’t that way it works. In reality there is a boundary (the Wall of Fame) defined by the equation
    Favorability Rating = 100%-Unfavorability Rating
    Any individual plotting along that boundary is known by all and all have an opinion regarding him or her. The region beyond that boundary is unattainable space.

    Hillary and Biden are well known and can’t get significantly more popular unless they change some minds. The GOP candidates are lesser known and struggle to gain favorability as they gain fame.

    The result is likely to be a negative campaign by the GOP candidate aimed at depressing turnout by Democrats and energizing the GOP base. It is the only formula that can work.

    • Matt McIrvin

      The 538 idea is basically to cast it in symmetrized and antisymmetrized coordinates: call them something like “fame” (favorable + unfavorable) and “net favorability” (favorable – unfavorable). Only they didn’t re-label the axes, they just rotated them, so the hard barrier at 100% fame wasn’t clearly noted on their diagram.

      I think you can see the Republicans’ problem on either version of the chart: net favorability gets more negative with increasing fame. And, of course, net favorability near zero doesn’t mean much when fame is low.

    • Michel Floyd

      The “Wall of Fame” wouldn’t be visible on the chart above. It intercepts the Y axis at 100% and the X axis at 100%. Only (50,50) is on the wall of fame.

      But I don’t understand why that is a problem. Neither the chart above nor 538s godawful confusing version of it model fame as moving along the line of equal favorability and unfavorability. The line is just drawn as a reference to give one an idea of net favorability (above the line positive, negative below)

      My bet is that as Ben Carson becomes better known the chart will have to be extended towards the right to accomate the (80,20) point. It would take Jesus reincarnated himself to get as high as (40,60).

    • Sam Wang

      I see, so Mr. Kettler wants a 45-45-90 triangle indicating all possible values. I agree, that would be better, especially if I have to plot someone in that pointy zone – like Bill Clinton, who is currently at an eye-popping 73-23. I think Bill is the secret weapon here.

      However, such a chart will cost extra. Oh, wait…this site is free! (insert winky emoticon here) Hmmm, maybe my pals Amanda Cox, Bill Marsh, and Co. at the NYT can cook it up sometime…

  • Eric

    I’m glad you included the bit about fitting a “Republican” line to all the GOP candidates. It would be interesting to see this repeated with a lot more Republicans and Democrats. Like, say, 20 or 30 of the best-known names from each party.

    • Sam Wang

      I was thinking the same thing. At some level, this implies that the right way to evaluate a specific candidate is to ask whether he/she is above or below that fit line. In other words, is he/she overperforming or underperforming the party as a whole?

  • Roger Moore

    What seems more obvious from the re-plotted graph is that none of the Republicans has an absolute favorability over 35%. My reading of that is that we’re basically seeing pure partisanship, and the Democrats have a partisan advantage.

    • Sam Wang

      Funny, I just added some text to indicate that. Thanks for noticing that.

      I note that whether one reaches this kind of insight is highly dependent on how one plots the data. Which argues a little bit against getting too clever, at least at the start.

  • Matt McIrvin

    …The other thing to consider is trends over time: the supporters of O’Malley and forlorn hopers for a Warren run like to talk about how Hillary Clinton’s favorables are dropping rapidly, and assume that will continue, rather than just being a return to ~50% because she’s turning back into a partisan candidate.

  • Matt McIrvin

    …The more I think about this election cycle, the more of a historical anomaly it seems to be. The expected pattern, the Reagan/Bush pattern perhaps, would have been to do it the other way around: run Hillary Clinton in 2008 and 2012, Barack Obama as her successor in 2016. (And, indeed, I recall some talk from Clinton’s supporters during the ’08 primary about how that was the natural order of things. But elections don’t work that way.)

  • David T

    Most of the Republicans aren’t very well known, but the people who know them aren’t a cross section of voters – I would imagine that they would trend toward conservatives (For example, Ben Carson is likely not known outside of the Fox News viewership)

    As a result, I would imagine this is an apples/oranges comparison, and when a more representative voter pool gets to know their positions, their favorability will drop as the sample isn’t skewed toward those who share their positions, no?

  • Matt McIrvin

    It is kind of interesting that nobody at all is in the triangle corresponding to actual majority popularity, up off the top of this graph.

  • MAT

    The other interesting point is that as high as Clinton’s favorability rating is to the R’s you’ve plotted, Obama’s, in the 7th year of his term, is even higher.