One of my favorite Zaphod quotes applies to this election cycle.

Put your analyst on danger money, baby

I don't remember that quote, but I put in the post on general faith.

After all, a man named "Bud Tugley" must be reliable. Only the name "Doc" inspires more confidence.

On the face of it this polling is encouraging because it asks people about a fait accompli, not about an inclination that may never be realized if someone doesn't bother voting.

The really funny thing is what that Zaphod line is in response to:

ZAPHOD:
Hey, “left here”? We’ve only just arrived.

GARKBIT:
Indeed sir. But before you arrived here sir, you left here.

ZAPHOD:
You’re saying that before we arrived here, we left here?

GARKBIT:
That is what I said sir.

ZAPHOD:
Put your analyst on danger money baby, now.

And you can trust me!

Ok Ok

want some good news?

this election has driven me to utter distraction. i am stunned, flabergasted, irritated and not a wee bit peeved that there are so many americans voting for kerry.

i just do not get it.

but, i could have lived with it and gone on my merry little way.

however, the pure vile pouring out of the left wing in this country combined with the complete and utter misunderstanding of realpolitik capped with the possibility of a virtual turncoat running this country has made me angry.

i voted for reagen in 84. i have never, since that date, voted either republican nor democrat - for any office. ever. and i vote.

this last little turn of events with the ammo dump, cbs, nyt and dnc colluding has infuriated me.

i was going to vote for bush and just leave it at that.

but now, i am volunteering my time all day thursday, friday, saturday, monday and tuesday to campaign in whatever capacity for the rnc.

i will drive people, walk door to door, call people, beg people and argue points day in and day out starting thursday.

i would like to give my thanks to the dnc, cbs, nyt, moveon.org and other nefarious organizations for energizing me to act more proactively - for bush

punk boy

Anyone who quotes TRATEOTU is ok in my book. Then again, you're talking to a guy who's computer is named Zaphod, whos laptops are Random and Trillian, printer - Marvin, router - Eddie, server - Ford, iPaq - Fenchurch.....

You get the idea.

punk boy I hear ya. I've never volunteered for a campaign in my life but I've put in three 4 hour shifts since Sunday and I've blanked out my calendar to work a straight Thursday thru Tuesday. Soros can kiss my ass before I give up my little chunk of Florida to him.

punk boy - welcome to the club. Already did one stint at the Republican HQ on the phone bank - and I normally would rather have a root canal than talk to strangers on the phone. But I don't want to look back afterwards and know I could have done more...

By the way - don't panic

A quick primer on why this poll (and pretty much every other poll) is unadulterated bullshit:

That famous "margin of error" arises from the formula for the confidence interval for a proportion...

95% CI = p +/- M.O.E.

= p +/- 1.96 * sqrt[(p(1-p))/n]

where p is the sample estimate of the true proportion, n is the sample size and "sqrt" means the square root function. Let's say you had a fair coin (true p=0.5) and flipped it 100 times. It might come out 50-50, but there's a reasonable chance it'll come out 48-52 or 56-44 or so on, even if the coin is fair. Now flip it 100 more times; you might get a different ratio of heads-to-tails. That formula above says in 95% of such coin flip experiments, the TRUE proportion is within the sample's margin of error.

Okay, so let's take that formula and simplify it; round off 1.96 to 2 (close enough for govt. work) and assume p = 0.5. The formula boils down to...

MOE = 1/sqrt(n).

In other words, the margin of error is simply one over the square root of sample size; if n=900, MOE = 1/30 = 0.033, if n = 2500 then MOE = 1/50 = 0.02. If you want to cut the MOE in half, you have to quadruple the sample size.

With that in mind, let's give this poll a statistical rectal exam. According to them, n was around 2400 registered voters, 1600 likelies; according to their results, 9% have already voted. Give 'em the benefit of the doubt and assume it was 9% of the 2400, or 216 total, who said they voted. Parenthetically, the MOE for this proportion is

1.96 * sqrt [(.09 * .91) / 2400 ] = 0.012

So, there's a 95% chance -- assuming that this sample was chosen randomly -- that between 7.8% and 10.2% of voters have already voted.

Here's where it starts to get stupid. Let's look at that subsample of 216 self-identified early voters. 51% (110) said they voted for Bush, 47% (102) said they voted for Kerry. The MOE for this smaller group is

1.96 * sqrt[(.51 * .49)/ 216] = 6.8%

Yeesh. Plus or minus sixteen point eight percent. Meaningless.

But here's where it really, really gets stupid: all of the above conjecture assumes a randomly selected sample. Product defects, lottery balls, coin flips.

What happens when, in order to flip a coin, you have to:

1. call the coin on its phone (land line only).
2. catch the coin at home.
3. make it past the coin's caller ID.
4. convince the coin to let you flip him.

And let's say that 65% of coins tell you to go fuck off. Now what's the "margin of error"?

MOE ~= 4.38528 * (length of three legged goat entrail) / log(circumference of Zogby's bald spot in furlongs)

Oops, make that "six point eight percent." Still meaningless.

The other concern I have with all the polls I've seen so far is that none of them have polled dead voters. ...They don't answer the phone, but it seems they usually vote democratic.

Plus, of course, the Republican coins tend to have jobs and/or tell the pollsters what to do with themselves, requiring yet another level of fudging.

I estimate that there's an uncorrected 2 point Democrat bias from that - so that all the polls are skewed against Bush by 2 points on average. That means the spread is 4 points wider in favour of Bush than the polls indicate. So if we're now seeing something like 52 - 46, it's going to be 54 - 44 on the day. We'll see soon enough.

If you read the ABC News article carefully you'll see that the voters they're talking about voting early are primarily out West. They make a special note of mentioning Oregon which as we know isn't exactly the heart of Bush Country, so fear not about Bush not being ahead by enough...We'd have to know more about where those early voters live to have any idea what relevance it might have to the overall vote.

Even though Georgia is not a battleground state...

In 2000, the election here went:
+Bush/Cheney 54.68
Gore/Lieberman 42.99

A good indicator for this election - one of the counties here anticipated some 9,000 early voters IN TOTAL and ordered ballots accordingly. As of today (Tuesday), they have already run out of ballots. Since there really has not been much of a Dem 'get out the vote' effort (nor Repub., for that matter!), I'm pretty sure this is really good news for the Prez.

oops, there should be % signs after those numbers. Lest you think the land of Jimmy is really so benighted that only 97+ actual people voted last election. Which, given that whole Jimmy thing you could not be entirely blamed for suspecting.

eaglespeak: By the way - don't panic

That's the first helpful or intelligible thing anybody's said to me all day.