Political Reporters, Polls, and Numerical Illiteracy

As Jesse at Pandagon notes, even though the presidential race is stagnant in that the numbers aren't shifting much, that's not the same as the race being in a dead heat. This would be obvious, if your typical political reporter wasn't a mathematically illiterate moron.

Yes, I know that there's only one poll that matters and it's in November, blah, blah, blah. But based on the available poll data, there's no way this is a dead heat. Why? Because, in 40 out of 41 polls since Clinton dropped out of the race, Obama has led McCain. If they truly were in a dead heat (i.e., 50/50), the probability that McCain would lead in fewer than two polls out of 41 is 0.00000000190993887372% (for those of you who like p-values, p = 0.0000000000190993887372).

To break the p = 0.05 threshold--that is, for the probability that McCain would lead in less than two out of 41 polls by chance to exceed five percent, Obama would be leading McCain 89% to 11% (even with that lopsided result, there would be a 94.9% chance that this observation is not due to randomness).

Someday, we will have a political press corps that is mathematically literate. And then I will have a magic pony.

More like this

Huh? How would a 'political press corps' be any kind of literate? They are political. Facty, not factual. Truthy, not truthful. And knowy, not knowledgeable. Why do you need them when you have Colbert?

By F'tang F'tang (not verified) on 12 Aug 2008 #permalink

What is even funnier is that they are using an indirect measure to begin with. Look at the state by state polls, tabulate electoral college predictions. Do some actual research and analysis.

Heaven forbid we remind people we don't directly elect a president but instead elect a slate of electors from our home state who in turn vote for a president. States got to have rights you know to protect those living in the pristine "rural" part of this country from those "urban dwellers".

Of course given the analysis (or lack thereof) that you mention in your post, that would require a lot more math and who has time for that after all we "need" our coverage of an affair of a FORMER Senator/FORMER presidential candidate, and talkingheads need to contribute their part to global warming. Sigh.

Yeah, but if they CALL it a "dead heat" instead, they have a better chance of convincing people that they need to keep consulting the media industry regularly because they think things could change at any moment.

I mean, as we saw in the last months of the Hillary Clinton/Barack Obama run-off, the news media desperately wants to avoid any hint of "game over" before the election, since that would mean they'd have to put in the effort to find something else to talk about to attract "eyeballs".

Not that I'm cynical or anything...

I always suspected something like that; that the fact that Obama is consistantly ahead (though not by much) of McCain is an indication of something more than just poll uncertainty. (If it were truly a dead heat you'd see the two candidates randomly changing from 1st to 2nd repsectivly. )

I was cautious to state that since I wasn't confident in my own understanding in the math and didn't want to jump to an incorrect conclusion.

As a foreign national, a resident 'alien' unable to vote, I have a hopefully less partisan view of the presidential race.

caveat: my Scottish upbringing and education persuades me to be somewhat parsimonious, and somewhat socialist in my viewpoint.

From my perspective, the media *is* hopelessly biased. Almost any evening I watch a news/analysis show (on whatever network) or read an article in a mainstream newspaper, I see unqualified reporting of republican memes (Obama is a fickle celebrity, McCain is steady & resolute; Obama is a free-spending, tax-hungry big government guy, McCain is tax cuts and fiscal responsibility, Obama is untested in foreign policy, McCain is deeply experienced)

My problem with the meme-replication is that most folks don't seem to actually look at the news. Any casual observation of the 'stories of the day' will show these memes for the blatant falsehoods they truly are. Unfortunately, when the story is played, the commentary includes whatever meme can be shoehorned in to fit.

Your math simply and numerically demonstrates what I've been observing. Obama is quantifiably ahead. But that does not fit with any of the existing biases, so it must be a dead heat, and must be due to his celebrity, not anything real.

fah!

tony:
Those of us citizens who are not hopelessly partisan Republicans largely agree with you. Usually it is attributed to the ownership of media by big business elites. I give more weight to the ownership of the advertising revenues which they depend upon, than to the actual ownership of the media itself, however the result is the same.

That said their is some validity to statements that the election is very close. If we had a popularly elected president, and the polls perfectly measured things (I'm talking about avoiding biases, not small sample size effects), and the electorate was stable, Mike's statistics would be valid. The problem is that there is likely systematic bias in the polls. And also a significant chance that one campaign will be able to shift the positions of a significant number of voters. Given those later systemic uncertainties, a statement of too close to call seems to be justified at this point in time.

You're being too harsh; the polls are measuring a moving target. It's not 41 samples of a static constant. For much of that time period Obama has been solidly in the lead. The gap has recently narrowed to the point where I think it is fair to say the race is now quite close.

After all, when looking at global warming it's the recent trend that's of interest - not the average of the "polls" of global temperature over the last 300 years.

A sobering thought (for those of us who believe that having a Republican in power is comparable to a Katrina that lasts for years, instead of days) is that in 1980, polls said that Carter versus Reagan was too close to call. In
1988, Dukakis led Bush in the polls until August 23. In 2004, Kerry was ahead of Bush until August. A few percentage points advantage in a poll really doesn't say anything about who might win in November.

It's these damn "independents" that making polling practically useless. If someone hasn't decided by now whether he wants to elect another Republican as President, then he is hopelessly ill-informed. Unfortunately, the upcoming election will be decided by exactly that group: the least informed, the shallowest, the most strongly influenced by trivialities.