Posted on | October 26, 2012 | 28 Comments
I am, according to some worthless putz named Joshua Holland writing at the left-wing site AlterNet. Holland also:
- Calls me “a notably dense right-wing blogger”; and
- Misspells my name (there is no “e” in Stacy).
As to the substance of the dispute, Holland’s article is entitled:
No, Republicans are actually (a) looking at Romney’s lead in numerous polls, (b) looking at Nate Silver’s bizarre transmutation of this data into a 73.1% likelihood of Obama’s re-election, and (c) saying, “What the f–k is Nate talking about?”
As been repeatedly pointed out, when you start digging down into the samples of the polls, there is good reason to believe that Romney’s vote is actually being underestimated. For example, it was big news that ABC/WaPo had Romney leading Obama 50-47. But Ace of Spades points out that the sample is D+4 and Romney leads by 19 points — NINETEEN FREAKING POINTS — among independents.
Let’s consider the possibility — just throwing this out there — that many of the “independents” leaning so decisively toward Romney are a bunch of bitter gun-clingers in Appalachia who don’t really matter in terms of the Crucial Battleground States.
Fine. OK. But excuse me for believing that if Romney has even half that big of a lead (9 or 10 points) among independents in Ohio, Florida, Colorado, Ohio, Iowa and New Hampshire, we’re gonna be looking at an all-out Republican romp Nov. 6.
Not even Jim Moran’s kid can steal enough votes to win it for Obama if Romney’s leading by 10 points among independents.
Back to the question, then: What the f–k is Nate Silver talking about? A commenter on yesterday’s post suggested it’s non-random “weighting”:
I found out everything I needed to know about Silver’s “analysis” when I learned that he was overweighting a week-old PPP poll that showed Obama well ahead and underweighting a fresh poll *by the same pollster* that had Obama’s lead almost gone in his “model.” The adjustments were not small. So he was purposefully and consciously giving a large amount of additional weight to old data compared to new data despite the source of both being the same.
This statistical legerdemain is justified by an ipse dixit assertion that PPP’s Dem lean has diminished over the cycle (a change he would have us believe has significant, observable effects over a 7-day period). Then there’s the grotesquely Obama-positive Marist poll to which he gave a huge statistical bump compared to a slew of more recent data for almost two weeks. That poll had a D+11 sample–over twice Obama’s 2008 ID edge and therefore something no rational poll observer could possibly argue with a straight face was likely to be valid. Nate loved it, though, giving it massively outsized influence over his overall results long after it was stale data.
Live by the weighting, die by the weighting, and if it weren’t for the fact that Nate is running this shady scam on the New York Times platform, we could laugh it off as irrelevant cheerleading. Instead, it’s highly relevant cheerleading, which has the effect of shaping elite opinion, which in turn affects media coverage, and thus tends — as a certain other statistical wizard once said — to “hide the decline.”
FLOUNDER: “Will that work?”
OTTER: “It’s gotta work better than the truth.”
Let me point out something: Nate Silver considers Rasmussen disreputable and unreliable, and yet on Thursday we could see three other national polls — ABC/WaPo, Gallup and Monmouth — all showing the same +3 margin for Romney that Rasmussen shows. How is it that these other polling firms report the exact same margin as the No Good Very Bad Stinky Rasmussen Poll Nate Silver Hates?
There is the possibility that Nate’s right and all of us “Nate Silver Truthers” are wrong. And there is the possibility that, even if Silver is wrong, his statistical cheerleading will have the self-fufilling prophecy effect of creating the phenomenon Silver pretends to observe.
On the other hand, there is also the possibility — and I think Nate Silver is severely underestimating this scenario — that he is so completely wrong his credibility will be destroyed on Nov. 6.
If Obama loses, we will not forget that he gave Obama’s likelihood of re-election at a precise 73.1% just ten days before Election Day.
At this point, Nate might as well raise the number to 100% — a guaranteed Obama win — and be done with it, because even if he tries to walk it back now, 73.1% is nearly 3-to-1, and if you lose a bet you called 3-to-1 with 10 days to go, your credibility is permanently screwed.
- Oct. 25: Polls Continue to Show Trend Toward Romney — Nate Silver Notwithstanding
- Oct. 23: Signs and Omens: Obama’s Fading Hope and the Graveyard Whistling Choir
- Oct. 22: The Democrat Graveyard Whistling Choir Ignores ‘Preference Cascade’ Scenario
- Oct. 20: My Beer vs. Nate Silver’s Two Beers
- Oct. 19: Nate Silver Asks: Whose Shark Is This, and Why Do I Feel a Need to Jump It?
- Oct. 18: GALLUP: ROMNEY 52, OBAMA 45 — Let the Great 2012 Liberal Freak-Out Begin!
- Oct. 15: Did You Say ‘Preference Cascade’?
- Oct. 14: ‘Media Credibility Day Is Coming’
- Oct. 11: Benghazi Breakdown: Team Obama, Liberals Now in Full Freak-Out Mode
- Oct. 9: Expect the Unexpected: Why Liberals Suddenly Melted Down After the Debate
- Oct. 8: Liberals Beginning to Realize They’ve Overestimated Obama’s Popularity?