Home 2013 races How Pathetically Bad Were Pollsters on Virginia 2013? Really, Really Bad.

How Pathetically Bad Were Pollsters on Virginia 2013? Really, Really Bad.

209
8
SHARE

According to UVA Political Science Professor @LarrySabato: “We need a fair, nonpartisan examination of VA GOV ’13 polling. It was off–in some cases (WaPo), WAY off.” It’s hard to disagree with that sentiment, given how utterly wrong the Virginia 2013 polls were in the closing weeks of the election. Click on the image to “embiggen,” or click here for the Huffington Post Pollster numbers for Virginia 2013. In short, they sucked. How bad did they suck? Let’s rank them in descending order of suckitude (based on how close they came to the actual outcome). Keep in mind that Terry McAuliffe ended up winning this race by 2.5 points (47.7%-45.2%).

1. Republican Rasmussen Reports, of course, was putrid as always, claiming on 10/20 that Terry McAuliffe had a 17-point lead. The only people who believed THAT poll probably also believe in little green men and unicorns and stuff.

2. Roanoke College’s “polling” unit, as always, embarrassed its school with its beyond-amateurish product. In a “poll” conducted from 10/21 to 10/27, Roanoke somehow managed to put McAuliffe up 15 points. Again, we’re talking little green men and unicorns. But what else is new for this bunch and its pathetically, laughably bad “polls” that for some reason “newspapers” keep citing as if they have any validity whatsoever? GIVE IT UP!

3. Our friends at the Kaplan/Bezos/Washington Post love to criticize automated polling and others that don’t meet their lofty standards, yet they (along with their partner Abt-SRBI) produced a total stinker of a “poll” in late October that had McAuliffe up 12 points. Why am I not surprised?

4. How is Zogby still in business, and why would anyone in their right mind take them even the least bit seriously when they pair up with the deranged “Newsmax?” This combo did one of the last (11/1-11/3) “polls” of the cycle, and also one of the worst, putting McAuliffe up 12 points. Again, how is Zogby still in business?

5. I probably should put NBC/Marist here, as they had McAuliffe up 8 points, BUT that was at the height of the government shutdown. Also, as far as I know Marist does decent work. So we’ll give them a partial pass, even though they were also off by a wide margin.

6. Public Policy Polling had the last (or next to last, it’s hard to keep them straight) poll of the cycle (11/2-11/3), and it was badly off, showing McAuliffe up 7 points, and indicating no reason whatsoever for any concern that this race was tightening, might be close, etc. I mean, why even bother doing polls if they’re going to be so misleading in terms of their absolute numbers as well as their analysis?

7. Christopher Newport University had a 7-point McAuliffe lead in its 10/25-10/30 poll. Not.

8. Quinnipiac and Hampton University each showed McAuliffe up 6 points in their final polls of the cycle. Actually, Quinnipiac would have been smarter to just stick with its 4-point margin of its next-to-last poll. Anyway, these two were among the “best of the worst” pollsters of Virginia this cycle. Which ain’t sayin’ much.

9. Now it’s where things get REALLY bizarre. First, an “Emerson College Polling Society” (a what?!?) “poll” actually nailed the results, at McAuliffe +2 points, but TOTALLY for the wrong reasons. Among other problems with that “poll”: they had Sarvis at 13 points (off by a mile); their internals were bizarre (e.g., they claimed that McAuliffe’s strongest support was in the 8th, 11th…and 10th (!!!!) CDs. WTF? The 10th?!? Uh, don’t think so. And how about the 3rd CD, which is overwhelmingly Democratic? This poll just proves that you can be “right for the wrong reasons,” also that you can hit a target merely by throwing darts at a board and getting lucky. Which brings us to the right-wingnut “Wenzel Strategies” (“poll” done for the Republican “Ending Spending” PAC). Wenzel is beyond parody, not even worth mentioning as a “pollster,” are wildly wrong ALL THE TIME, yet somehow they came pretty close in this race, showing McAuliffe up 1 point. Who knows, but in no way, shape or form should anyone EVER take this group seriously in its future “polls.”

So, yeah, Larry Sabato’s right that “We need a fair, nonpartisan examination of VA GOV ’13 polling.” But I’ll bet you a lot of money that we won’t get one. I hope I’m wrong, but I’m not holding my breath.

P.S. In fairness, none of the pollsters found Cuccinelli ahead, and they were right about that – barely!

  • unstablefan

    All the coverage of Terry having a huge lead may have made D voters complacent. I’m not sure that the pre-election polling was THAT far off.

    However, the results also do not align at all with the demographic information in the exit polls, which I find more surprising and perhaps troubling.

  • FreeDem

    Terry’s polling had him up only 3. Their modeling was accurate.

  • mlemmons58

    The hardest thing about this kind of polling is figuring out who’s going to bother to vote. It’s difficult in a Presidential year, and exponentially worse in an off-off year like this one. If that examination ever happens, it would be nice if it included a look at how much of the margin of error was due to getting the likely voter screen wrong.

  • Andy Schmookler

    True.  The polls said X would happen (McA in decisive fashion, if not a landslide) and Y happened instead (fairly narrow victory).  Maybe that means that the polls had it wrong.  Maybe not.  I’m betting not.

    There are two important things that changed as the election grew closer. First, the shutdown started receding into the past and the debacle with healthcare.gov was giving the Democrats a black eye. The Democratic moment hit a crest back at shutdown time, and the tide has been running the other way for a while.

    Also, I am hearing that there was an unexpected surge toward voting on the Republican side. For whatever reason, they rallied and voted in numbers beyond what was anticipated. So I’ve heard. (Am I wrong.)

    So the pollsters may well have read the temperature right for the moment they took the polls.  But the weather changed.