Wednesday, September 19, 2012

Beware the Poll Write-Ups!

Oy, Pew:
At this stage in the campaign, Barack Obama is in a strong position compared with past victorious presidential candidates. With an eight-point lead over Mitt Romney among likely voters, Obama holds a bigger September lead than the last three candidates who went on to win in November, including Obama four years ago. In elections since 1988, only Bill Clinton, in 1992 and 1996, entered the fall with a larger advantage.
That Barack Obama was winning by 8 points last week -- the poll was taken September 12-16 and released today -- is a pretty good reading for him.

But, c'mon. This poll was done very soon after Obama's convention (September 4-6), when his bounce was still in effect. Now, let's see about the "last three candidates" Pew is comparing Obama 2012 to. In 2008, they had Obama exactly tied with John McCain...but that one (seems to have been September 9-14) would have been during McCain's convention bounce; the Republicans went last in 2008, with their convention taking place September 1-4. In 2004, the Republicans met August 30 - September 2. The one in their current table was taken September 11-14 and had Bush up one point...but they also fielded a September 8-10 survey, and that had Bush up a whopping 16 points! Plus a September 17-21 poll had Bush +3. Going back to 2000, the Democrats met in mid-August, well ahead of Pew's September poll that had Al Gore up by 5, so that one is at least unrelated to a convention bounce.

Basically, they're overselling Obama's advantage, because they're not comparing like with like. The 2004 case is particularly bad, but overall the advantage for Obama here is just a lot less impressive than the write-up would lead us to believe.

Moreover: since all individual polls are subject to random variation, this kind of comparison just multiplies that kind of error. Think about the three September 2004 polls: odds are that Pew was "wrong" high on the first one, low on the second, and about right on the third (even given that Bush's read lead was probably higher early in the month than later).

Again: any 8 point lead in a poll is good news, and I'll agree that a September lead is a lot more meaningful than a July lead. The polling averages suggest that Obama's lead last week was more like 3 or 4 points than 8 (and no, it wouldn't totally discredit a pollster to include that in the write-up, at least not in my view), but sure, even post-convention a candidate is still going to be happier up by a few than even or down. The problem here is just that the added comparison to 2008, 2004, and 2000 confuses more than it clarifies -- and yet it's highlighted in their release -- what I quoted above is the top paragraph of the write-up on the web page.

Pew does great work, as most of the nationally known pollsters do. What this example reminds us, however, is that the write-ups are often just as prone to error as the numbers -- and, perhaps, even more.

21 comments:

  1. What the hell is your point?????

    ReplyDelete
    Replies
    1. The point is pretty clear - it's in the paragraph starting with "Basically"

      Delete
    2. To be fair: that is the 5th paragraph, more or less.

      Delete
  2. Well, one could argue that Obama's convention bounce had totally receded before Pew's poll was in the field. According to RCP, Gallup had Obama up 7 in the week of 9/5-9/11; that included the last two days of the DNC. The next week, though, 9/12-9/18, they have Obama up by only 1. It would be reasonable to assume that, if the post-convention bounce ended on a particular date, 9/11 might have been that day (especially considering that was the day of the Libya murders). Pew as 9/12-9/16.

    But your broader point is of course correct. Frankly, I think poll write-ups, as a whole, are useless. Thankfully, we have poll aggregators to dampen outliers and show us the true state of the race (or at least the state of the polls).

    ReplyDelete
    Replies
    1. Incidentally, Andrew Sullivan just made the same mistake as Pew is accused of in this post. (Even worse actually; he's comparing post-RNC Obama-McCain to post-DNC Obama-Romney)

      Delete
    2. Andrew Sullivan, as you probably know, writes about this stuff with his heart,not with his head. So it is even worse than most.

      Delete
  3. Jon's point is that this number is almost certainly inflated, if this survey was conducted too close to the Charlotte convention (when, as you know, Gallup in particular recorded a major bounce for Obama). Conceivably, Obama could win 56-43 or 57-42 IF the people that would vote for him actually do so, which would more than return the House & push the Dems close to 55 Senate seats. But how many will? Pew's finding that Obama's numbers are virtually identical between registered & likely voters is the real outlier. This essentially assumes it is 2008 all over again (minus the demoralized state of the GOP). Maybe so. People who voted for Obama aren't voting for Romney, but I'm willing to submit that a small sliver who voted for Mad Mac may vote for Obama this time. The real key of course is who would vote for just one 2012 candidate but won't show up after all. Romney's failure to relate w/ men is especially glaring here (even if he's likely up 5-10 as opposed to 1); he has to at least match Bush's 2004 performance there to have any chance, probably a bit more given his weaknesses w/ Hispanics. Dems should still be nervous about FL, OH, & WI until the end (perhaps less so VA), but if Romney's not leading in the average of the final polls, he's not winning. Unfortunately, Obama has to do more than win; he has to hit a margin that will take back the House. Hopefully Silver or someone else will tell us what number that would take. I would guess 52% is enough.

    ReplyDelete
    Replies
    1. No, 52% won't likely do it. Most of the prognosticators are calling the House a wash as of now, and I can't imagine a 4-6% margin would be enough to move that much.

      We're about 6-7 weeks out from the election, and it appears Obama doesn't dare show his face in any more than a few dozen House districts, and that's assuming they'll have him. I don't think you should plan on any House flipping this cycle.

      Delete
    2. Over at the Monkey Cage, they're predicting that the Democrats will gain only one seat in the House. At they same time, the numbers are fuzzy enough that they give them a 1 in 4 chance of taking the House back.

      Delete
  4. I think the constantly polling doesn't help anybody figure out what's going on in terms of what it'll mean in November. The debates will say more and Romney now must defend what his right wing base has always wanted-- condemning those needing government help as parasites (Mary Matalin) and doubling down on the Ayn Rand philosophy. That will lead to cheers by his right wing voters who would vote for him anyway but how about the middle where they are being asked to dismantle everything they have grown up believing helps us as a people from public education to a safety net. His debate performance will have to deal with also what he said about Palestinians not wanting peace. He can't back up given he's been called an etch-a-sketch and what he has to defend is going to go contrary to a lot in the middle. This thing of thinking redistribution will help him, ignores that all taxation is redistribution to some degree. Does he truly want to defend a dog eat dog world? It won't make his debate performances that easy but he'll have cheers from the Fox crew no matter what.

    ReplyDelete
  5. Yeah, this Pew poll is bogus. I can see Obama matching his 2008 performance, more or less, but I can't see him opening it up to 10-12 points, as an 8 point lead today surely implies to be the final result. I'd definitely bet the Under on Obama's 2008 totals, this time through.

    Obama is the luckiest man alive, drawing Willard as his opponent this cycle.

    ReplyDelete
    Replies
    1. Or Obama got to follow GWB. Since the revolving door 1888-96 presidential elections, only once has a party in the White House been bounced after only four years & it took a lot in 1980 to oust Carter. Obama wasn't challenged in the primary season, did not have anything close to an Iran hostage crisis, & presided over a far less significant "misery index," I'm not sure any Republican could have beaten him this time. And the only ones who could have could not have been nominated by this current GOP. Watching OH on Election Night will be interesting because at the moment Obama is clearly overperforming there. If that state is turning into MI then the Republican route to 270 is all the less plausible.

      Delete
    2. I can't see Ohio as anything special. It appears to be performing to type. I don't think the Midwest likes Willard very much, is his major problem. Yeah, I'd carry IA, WI and OH as tossups right now, but I think he's going to eventually lose big throughout, only taking IN at the end of the day.

      I think that's where the R's screwed up. A more Midwest oriented candidate would have hurt Obama. Willard is nothing close to that.

      Delete
    3. I can't see him opening it up to 10-12 points, as an 8 point lead today surely implies to be the final result.

      Why would a 8-point lead now translate to 10-12 in November? The main point of this post is that Obama might still be coming off his convention bounce. If anything, an 8-point lead now should translate into a smaller margin in November.

      Delete
    4. Generally speaking, incumbent presidents either split or win the undecided vote, so an 8 point margin now would most often open up wider later on, meaning a 10-12 point margin.

      And I can't see this race going to a 10-12 point margin, so that makes the Pew poll bogus.

      Obama's ceiling is the 2008 margin. And again, he's lucky that Willard is his opponent, as that guy is his own ceiling.

      Delete
  6. To extend your point: PPP's analysis of its own polling often strikes me as just terrible. The sort of thing you use as examples of what not to do.

    One example: Obama and Romney with negative approval/favorability numbers respectively with the Ohio electorate indicates an electorate "disgusted with its serious choices".

    I did some arithmetic and figured out that about 3/4 of their respondents--at least--either approved of Obama's performance or had a favorable opinion of Mitt Romney.

    ReplyDelete
    Replies
    1. And again, last week!

      "Independents in Minnesota- Obama approval 45/52, Romney favorability 40/51. Good prism into unhappiness of electorate"

      http://twitter.com/ppppolls/status/245980702447919104

      No, you know what would be a "good prism into unhappiness of electorate"? Asking people "are you happy or unhappy with your choices?". How they'd answer that is besides the point, but PPP's numbers don't really mean that, and the number of Minnesota independents they polled who actually disapprove of Obama and have an unfavorable opinion of Romney is probably tiny.

      Delete
  7. For what it's worth, here's Gallup with a fair WRITEUP but a lousy...just terrible..headline and lede.

    http://www.gallup.com/poll/157544/voters-reaction-romney-comments-tilts-negative.aspx

    The writeup admits "OK, this doesn't matter, but....", whereas the lede and headline are shoveling dirt on Romney's grave.

    ReplyDelete
  8. I really can't stand those "will it make you more or less likely to vote for?" questions. I mean, more than is rational, which would still be quite a bit.

    There's just no way that the answers have any relationship at all to the surface meaning, so why ask it like that in the first place?

    Which is to say that I couldn't get past what the question was to figure out whether they wrote it up well.

    ReplyDelete
    Replies
    1. Honestly, the stupidity of the question just jumped out at me, so I clicked through. (I also REALLY didn't want deal with personnel committee work)

      Delete
  9. Polls are more for the politician than they are the voters. The problem with polls is that it is very hard to figure out who was polled. Polls are, at times, to provide a favorable or unfavorable poll of a particular question or issue.

    If the politician is doing bad, and wants it to look like they are doing better, they poll their base. Democrats and Republicans both do this. When a poll is done by the media, it typical polls they demographic that favors that media's views.

    Like you said, by adding the prior years to the poll only confuses the true numbers. People read this as Obama is doing better than everyone except Clinton. The common person seeing this on the nightly news doesn't hear about which party's convention went first on those years.

    ReplyDelete

Note: Only a member of this blog may post a comment.