Why the Polls Matter... And Why They Don't

Apropos this wonky little observation from Seth Colter Walls on the Huffington Post site:

http://www.huffingtonpost.com/2008/09/09/poll-madness-mccain-takes_n_125158.html

This gave me a big, fat, "ah-HAAA!!!" moment. I have been mystified by the polls reporting a McCain lead for a number of reasons, and by the apparent despondency among Obama supporters in response to those reports for one simple reason. Let me explain:

The polls have been mystifying me because as far as I can tell, the addition of Palin to the GOPpie ticket, for all the noise it's generated, doesn't seem to have actually added any likely voters to their prospects. And, based on what happened at the RNC, they are actually losing support in some key groups-- borne out by some anecdotal, but cognitively harmonious reports.

One thing that completely flabbergasted me about the RNC was their complete abandonment of an already large --and fast-growing-- voter bloc: America's diverse Latino/Chicano/Hispanic population. The GOP has spent a lot of energy wooing this population over the last twenty years, but at their convention it became clear that the GOPpies were happy to toss them in the crapper. The attempts of a few speakers to address a presumed Spanish-speaking television audience received boos and catcalls from the lily-white live audience who apparently objected to acknowledging the existence of millions of Americans.

They're not as reliably "solid" in their voting tendencies as other groups (a fact that gave the GOP a toehold among them to begin with,) but as the 2007 "white t-shirt" marches illustrated, once they are motivated, they take action. And one sure way to motivate them is to treat them as though they don't matter.

In the last few days I've taken some informal pulse, and listened to a good many friends and relatives who've been doing the same: With the exception of the most extreme fetus-worshippers, Sarah Palin's convention speech alienated a lot of Catholics. Subsequent utterances have only magnified the effect. And with the exception of the most extreme ultra-Zionists, she's worrying America's Jewish voters. Granted, the fetus-worshippers and the ultra-Zionists are a noisy bunch, but they're a minority-- and not even a numerically substantial one-- among each of those populations.

And then there are articles like this:
http://www.post-gazette.com/pg/08252/910390-470.stm

And editorials like this (In the KC Star, no less):
http://www.kansascity.com/276/story/787928.html

For all the hyperventilating hype about "women flocking" (like hens, presumably?) to the McCain ticket, it certainly doesn't seem to be a universal trend. I sat next to a die-hard GOPpie in the pharmacy yesterday, both of us waiting to pick up our prescriptions, who lamented that McCain had passed over "really qualified" women like Kay Bailey Hutchinson or Christine Todd Whitman for "a religious nut." She doesn't like the notion of McCain putting Palin a heartbeat from the Presidency any more than I do. She didn't come out and say she'd vote for Obama, but she admitted that she was "thinking about it."

And it's clear that the effect of the bravura DNC on the Democratic electorate was like dropping one last crystal into a supersaturated solution: It solidified. Traditional Democratic voter bases like Obama and will turn out to vote for him in record numbers. In fact, Democrats have added so many new registered voters to the rolls in the last few months that we now hold a solid advantage of 11 million more voters than the GOP.

So I'm already definitely looking sideways at the polls trumpeting a McCain groundswell. Say, what? Who are these people talking to?

Well. According to Mr. Walls' astute examination, they're being --to put it politely-- manipulated. It's a wonkily technical manipulation based on skewing the sampling, nothing so blatant as push-polling or suppressing responses, but it's nevertheless manipulation. Here's a quickie analogy to clarify:

Suppose the State High School Football Championship is being played in the capital, and you're trying to figure out how many fans of Lincoln High will attend, versus how many fans of Washington High. So you go ask a selected group of known high school football fans if they're planning on attending the game, and if so, who they'll be rooting for. Your sample of 100 fans indicates that 52 plan to attend and cheer on Washington, and 46 plan to come in Lincoln colors. Pretty conclusive?

Not really. Because in picking your sample you picked forty-five fans who'd supported Lincoln in the past, forty-five who'd supported Washington in the past, and ten who'd supported Truman High, knocked out of the playoffs in an early round by Lincoln. BUT-- and it's a rather large "but"-- Lincoln High is a large urban school right in the capital, with 1200 students. It's been around for fifty years and has many thousands of loyal alums who've been going hungry for a championship for nearly a decade. Washington, on the other hand, is in a small, wealthy suburb. It has only eight hundred and fifty students. They've won three championships in the last seven years.

If I were planning the ticket distribution in the 10,000 seat stadium, I sure wouldn't plan on only 4600 Lincoln fans showing up. And I'd be very skeptical of whether Washington would manage to get 5200 fans to the game.

That's why I think the polls don't matter.

And here's why they do matter:

Flawed polls: Cui bono?

I don't think the GOPpie strategists and their enablers among the media barons really expect to keep enough despondent Obama voters at home to throw the election to McCain. I don't think it could happen. The math doesn't work. I think they're hoping for two things, and both of them could happen:

1. They want to let a little air out of the sails of the boat that's been sailing rings around them. The Obama campaign runs on grassroots energy, optimism, and hope. Deflating that, even a little bit, could take some momentum out of our efforts to register new voters, raise grassroots money, get volunteers on the phone banks, and get first-time voters to the polls in November. It probably wouldn't be enough to let McCain win. But, combined with #2, it just might be enough to let them steal a few key districts in a few key states, and once again tip the balance.

2. To get away with a steal, they have to make the outcome look plausible. They can't have a long history of polls showing Obama holding, increasing, holding, holding, dipping, holding, increasing, holding, increasing, and holding a lead. They have to make it look close. They have to make a credible case for McCain to appear competitive. A lead now and then, even if he can't hang onto it, seesawing, staying close, pulling ahead, dropping back, holding, pulling even... THEN it becomes plausible that he could just "squeak out" a win in November.

And there you have it. If McCain wins, the big media operations don't have to go through all the expense and trouble of re-tooling, as they will if Obama wins (Obama victory: watch the media morph overnight into slavering, snarling, relentless anti-government watchdogs. But it would require some costly adjustments.) If McCain wins, the gravy train derails.

Cui bono.

So the polls matter, whether they're reliable or not. I'm not relying on them. I don't think they say squat about what's really going on in the minds and hearts of people who are really going to go to the polls in November. But they matter, all the same. We need to call the polling organizations on their little cheats, slap them down every time they advance the silly notion that there's a real "horse race" going on here, demand crosstabs, question every sampling procedure.

But NOT regard them as reflective of anything remotely resembling reality, and most emphatically NOT allow them to distract us from the task at hand: Restoring hope to America.

determinedly,
Bright

very subject and not getting my point across. Thats exactly what I have been trying to communicate to everyone, but who the hell pays any attention to someone with the word "Republican" in their pseudonym around here. While I do enjoy my rants here at DU, it gets depressing sometimes. When I offer up some insight and a point of view from the other side and am completely ignored because folks do not know what to make of my motives or they just think I am bat shit crazy because of my pseudonym. I do not have a flare for written expression. You have just summed up just what I have been trying to warn people about in a way that I could never do, thank you for your excellent penmanship er keyboardmanship, uh whatever, you get the idea.

:toast:

pseudonym after all, it just must be the subject. People just do not understand or we are just fringe thinkers.

:shrug:

Now, don't both campaigns pay big money for internal polls that don't manipulate the samples?

Don't these samples more accurately represent the actual proportions of the electorate and accurately reflect the voters' intentions?

If I were running for office, I know I'd want polls that told me what the electorate was *really* feeling, not what the media wanted me to think.

And so, if this is true, that the McCain and Obama campaigns have internal polls that more accurately reflect the sentiments of the electorate, what do we make of the campaigns actions in the past week, and the coming weeks? I understand they are trying to mold the narrative, just as the MSM is trying mold the narrative, so the public will believe it's a close race; Obama doesn't want to look cocky or overconfident, McCain doesn't want to appear desparate...anyone have any thoughts?

Thanks again for your analysis.

in both the polling and the tallies. K & R

Up until about today I was doing just fine. Then I started to panic. Thanks for the reality check. It all makes sense.

I feel like we're still on the rails of reality. In other words, I realize that they're not for real. But we are.


One other thing- the more their polling is manipulated to skew in favor of McCain, the more we pull together to get Obama elected. So in a way, they are their worst enemy.

So many smart people on the same page, and some wonderful, intelligent writers here.

http://journals.democraticunderground.com/Wetzelbill/377


I slightly updated it on my myspace blog with the Seth Colter Walls info. Basically it boils down to, McCain's lead in those polls with the faulty methodology isn't what it appears to be and this race is still close either way. However, the trends favor Obama, from the electoral college, to voter registration, potential turnout and even the general direction of the country and political climate. McCain may not have a lead at all in those polls if the methodology was accurate.

That being said polls are always faulty because they cannot truly and properly account for voter registration and turnout advantages. Sometimes they don't even try to account for those.

A page of links would be very helpful to follow up. Can an individual email or call to complain?

I was beginning to see this, but you put it into words.

...every day they write another post stating their gloom and doom. Polls should be taken with a huge grain of salt. They are one snapshot in time, and, as most statistical data, can be manipulated in many ways, to get whatever outcome one is seeking.

K&R