Thursday, October 18, 2012

ABC/WaPo Poll Should Be Ashamed of Themselves

On Monday, the Washington Post and ABC published a poll that claimed Barack Obama was up 3% over Mitt Romney in their latest poll. There's just one problem: They used a sample of +9% Democrat. (A full 2% above the +7% Democrat record turnout in 2008.)

Do you want to know what makes this publish so preposterous? Directly from the Washington Post's article reporting this story:

Partisan identification fluctuates from poll to poll as basic orientations shift and with the sampling variability that accompanies each randomly selected sample of voters. In the current poll, Democrats outnumber Republicans by nine percentage points among likely voters; the previous three Post-ABC polls had three-, six- and five-percentage-point edges for Democrats. The presidential contest would now be neck and neck nationally with any of these margins.

Friends, the Washington Post ADMITTED IT! They admitted this was a bad poll! So why, pray tell, did they publish it? Let's make pretend it was an accident. Lets pretend this was a sampling anomaly (just to be charitable). Why would they publish this poll? It was clearly misleading.
Now I can see you thinking "what would the numbers have been if they used a real believable sample? Let's give you some quick numbers, shall we?

With a +3% Democrat Sample, Romney is leading by 1%, 48% to 47%.

With a flat even sample, by the way (which I believe is very likely given the 2010 election turnout) it becomes a Romney lead of 49% to 46%. It's a complete switch.

Some might ask why this matters. Well, aside from the fact that the Washington Post stated that the poll had a misleading sample IN THEIR ARTICLE, let's look at our old friend the Real Clear Politics average. Since I have an established propensity to not publish posts until a few days later, lets show you what the RCP average was on Monday, October 15th (when I wrote this post):


Notice that the polls other than this anomalous ABC/Washington Post poll have Obama at best at +1% to Romney at +2%, giving an RCP average of Romney +0.1%. Now, let's remove the Washington Post poll entirely. Just throw it right out. That moves the average from Romney plus 0.1% to Romney +0.6%. Now when you unskew it to a +3% Democrat sample (and giving Romney a +2% lead) we get to an RCP average of Romney 47.9%, Obama 46.8% to give Romney a +1.1% lead.

So I say, as I said in the title, shame on ABC and the Washington Post. How dare they publish this admittedly biased poll. They could have weighted the poll to a more reasonable sample easily (as most polls do) if the problem was really just the sample. That's easy to do, gang. They could have done it with just a modicum of honesty.

Instead, the Washington Post chose to publish this poll, they published this link on Real Clear Politics.. If you went to the Washington Post politics page, you could find this article.  In the third paragraph, it said "Obama 49%, Romney 46%." It's not until the 8th paragraph do we see "you really can't trust this poll because the sample is silly."  Most people won't read to the 8th paragraph of a story. Some of us did.

The truth is the Washington Post should be ashamed of themselves for publishing this poll. That's the truth they won't tell you. This poll is at best a horrible mistake. At worst, it's a willful lie.

No comments:

Post a Comment

All posts will be reviewed subject to the Rules for Commenting. Any post that does not abide by these rules will not be posted, entirely at the discretion of the blog editor.

Commenters who repeatedly violate these rules will be permanently banned from commenting, and thus none of their comments, regardless of content, will be posted.