Wednesday, June 6, 2012

Calibration and Instrument accuracy.

Looking at the errors in the exit polls in Wisconsin versus the actual results, Roger L Simon wonders if the political class' measuring devices aren't all off:

Apparently, the silent majority of Wisconsin voters didn’t want to admit to nosy pollsters and anyone else that might be listening that they were opposed to runaway unions, runaway spending, or the Democratic administration. They just wanted to cast their votes. And they did.

This Bradley Effect, then, is not like the Bradley Effect of yore. It’s about race to some degree, but I suspect there are much larger components of being fed up with elites of all sorts, interest groups, media groups, union groups, all sorts of groups telling the average citizen what he should and shouldn’t think, openly or covertly threatening to ostracize him or her for not going along with the pervasive liberal status quo. This was a cry of “Ya, basta!”

So if I were a member of the Democratic Party this morning, if I were David Axelrod and his team of so-called wise men, I would be wondering – what if all the polls are wrong? What if this is true across the entire country?

What does it matter if their measurements are just a little bit off?

Even if these polls are wrong by three or four points in only a handful of states, the results of the coming election could be disastrous for the Democrats. Romney could win in a walk and bring a Republican House and Senate with him.


On a related note this is why I'm eyeing a better scale, specifically one with better resolution and consistency. I've got a numerical example that I'll post later that is quite interesting.

My scale at *best* gives an accuracy of +/- 0.05 grains.

Now consider that I'm working with standard deviations on that same order.

Such that a bunch of readings at 4.05 could go to 4.0 or 4.1 which would have a big influence on the statistics. To make it worse I certianly have more error than +/- 0.05 grains.

This is why I'm thinking of stabalizing on a load at (starting load + 0.2 grains).
Based on the spread I get with Unique that would give me a good hold where I'd be within the safe zone *and* would be biased towards the low end for an extra factor of safety.

This is of course after I've tested at (starting load + 0.0 grains)

Now consider that polling data is far, far more questionable than scale data.

Granted apples and oranges, but the point holds that you can only make judgement calls based on incoming data, and in many ways the accuracy of that data is more important than the data itself.

If I tell you the reading is 5 grains or up by 5% of the vote. Well that's nice, but you'll want some context to how good those numbers are.

In a previous comment Mr. B gave a bit of advice for those that want to straddle the "hot load line", use a stronger case. That's because even if you account for the spread in performance of the tool, what your using to measure might be wrong.

An example of the dangers of using incorrect but comforting data. The Obama campaign is taking basking in "exit polling [that] showed President Obama beating Mitt Romney 52-43,"
The problem is that said exit poling also showed Walker and Barrett being about 50-50 split.

When Walker actually won by 7.5 percent.

And when the exit poll results are scaled based on the actual Walker Barrett split the result is 48-48. Now such a scale is also fraught with risk, but one should not accept data that you know is bad and base a decision on that, especially if its the choice you wanted to make.

That's textbook confirmation bias. Also it's a bit funny to hear Wisconsin being talked about as if it were always a battleground state. Then again there's folks who think the South was always solid Republican or that there's never been a national democrat from Texas.

It all doesn't keep the Admin from saying: "no one can dispute the strong message sent to Governor Walker."

The message of when given the choice of Barrett versus Walker a second time the voters respond the same way, but now with more cowbell?

Unfortuantly the voters interfered wtih that "message" as Ace notes:

After the 9 7 point drubbing, team Obama wanted to talk about the "message" sent to Scott Walker by the public.

The message was apparently "Stop your right-wingnuttery or we'll vote for you in higher proportions than in 2010.

And the Obama admin still keeps up with the "how close this contest was" lie.

So, I think there's something to Roger L. Simon's theory.

And speaking of blitter data clingers. Some still insist that the exit poll data is correct, and that the real winner was Obama:
Tell you what Team Exit Polls...show me what a voter who supports Obama and Scott Walker looks like and then I'll take you seriously (actually, I'll laugh at them but close enough).

I'd love to see a Venn Diagram of people who believe exit poll data over actual votes and people who believe climate models over observable climate data. I bet there's a big overlap.

Though at least it's not as fantastical as the "Well the evil Koch bothers and Walker just bought the election." argument Ron Radosh demolishes this literally simplistic, repetative, and cartoonish argument.
These representatives of the people do not seem to comprehend how their argument shows little confidence in the ability of the average Joe- whom they claim to represent-can understand their own needs, and make wise decisions on that basis. If Dreier and Brown are right, they are in effect saying that the people are so dumb, that TV commercials paid for by the Koch brothers are enough to produce an outcome contrary to their own interests. (Dreier does not know, evidently, that the Koch brothers favor same sex marriage and decriminalization of marijuana, causes he probably supports himself. Maybe they are responsible for Obama’s sudden turn on that issue.)

...

"Or, the people are dumb. Yet, they do not make that argument when their side wins electoral contests, and billionaires like George Soros makes greater contributions than the Koch Brothers ever made to produce victories for leftists. They did not make that argument when a few years back, former Gov. John Corzine of New Jersey ran the most expensive election in that state’s history, spending his own fortune to assure his election. When that happens, they declare only that the people choose wisely.

Representational goverment is fine, as long as we all agree to have the right kind of representatives.

What the Dreier-Brown argument shows is how bankrupt the Left is in its ability to comprehend reality. As I and others pointed out earlier, Scott Walker has been a success in Wisconsin. He has lowered property taxes, saved teachers from being fired, exposed the shell game practiced by public sector unions, improved the economy and fiscal standing of his state, and generally proved to be a popular reform Governor.

But it can't be that. Only the right has ecochambers.

So that is there simple answer: The Republicans spent more. They can ignore everything else that stands in the way of their inability to see why they were defeated. Every left-wing site makes that point alone, and repeats it endlessly. They have no other explanation.

No wonder. As the Left sees things, they are correct and represent the real people. When those they claim to speak for vote against them, the only reaction they have is bewilderment. And so they come up with one answer their followers can understand—the rich bought the outcome.

A comforting lie doesn't help guys. Just look at the Republican party in... well any time. And of course Obama's right at the front of demonizing money in politics, right after 5 fundraisers in a day.


Meanwhile Jammie Wearing Fool reminds us that those whining about money in politics haven't been running this recall effort on a shoe string themselves.
The swift results left Democrats in deep denial, whining endlessly about too much money in politics. Yes, really. The folks who bought and paid for this recall, who for 16 months ran a campaign filled with lies, are now fretting over too much money being spent. You can’t make this up.
...
This was clearly the coordinated excuse Tuesday night, one we’ll hear nonstop from the media and Democrats in coming days. No, the 16-month temper tantrum and rejection of hard left unions has nothing to do with it. It’s all about the money. This was the direction MSNBC idiot Lawrence O’Donnell took, but first he provided a moment to cherish, declaring Obama the big winner last night. Yes, really.

And O'Donnell explicitly used said exit polls. Once again taking comfort in bad data and the convenient lie.

It's not like Big Labor funded over twenty million dollars in this. No just focus on poor Mayor Barrett. And its not like that union money came from dues, that were largely collected due to state mandate. Mandate that Walker played a part in removing with his reforms.

Reforms that when it made being a member and paying into a Public union voluntary saw membership decline by over half. Huh. But no, the only side with a vested fiscal interest here are the fat-cats. Err the republican fat-cats.


Back to Roger L. Simon, even if the polls were right. Would the Progs believe them?


@comradearthur made a good point: Scott Walker was only in office for two and a change years, rather than Obama's three and change years, but Scott Walker's reforms produced tangible, positive results in that time-frame, whereas Obama is still blibble-blabbing about "needing more time."

This is the acid test of politics: Do your policies correlate with positive results, or do they not? Theory and ideology and partisan rooting-interest are all well and good, but ultimately, do your policies actually produce the results you predict they will?


And Tam takes comfort that the result was outside the "margin of recount." And has good fun on echo chambers. I'll give a preview:

Well, the voice of the people was heard, alright, only it didn't say what the HuffPo crowd wanted it to say. Now they sulk that it really is "too hard to fire a rotten public sector employee," and go back to their online echo chambers and complain that the election must have been rigged, because not one person they knew was going to vote for Walker!

Hehe. Hard to fire. It's like when my Canadian friends complain about all the stuff Big-Bad Harper is doing, and I reply with: "Gee, its a shame the Prime Minster has so much power. Shame there aren't more checks on what the State can and cannot do."

No comments: