Saturday, September 10, 2005

Cheap opinions

Capital Freedom had very perspicacious thoughts on the contradictory answers people gave in a post-Katrina CNN poll. Among them: "If 56% believe that New Orleans is 'beyond repair,' meaning that it cannot be repaired, how can 63% be in favor of rebuilding it? Perhaps the definition of 'beyond repair' is a little vague." Indeed, it's too hard to account for what people really mean, which is why I rarely pay attention to polls. She concluded, "The conclusions are virtually meaningless without background on why the respondents chose the responses they did and what they meant by their responses. We can neither explain nor understand the responses to polls like this one without also knowing the reasoning that holds the responses together."

I question how these "609 adults" are so qualified that their opinions are important enough to be recognized in a poll. That's not a personal insult, merely an objective observation that these people may not know much at all about the issue. Are they New Orleans homeowners, who would know better than anyone whether it's better to rebuild or replace? Are they contractors who are experienced and expert with hurricane and flood damage? Or are they people like me who do not live in the area, have not seen the destruction first-hand, and thus have only the most superficial knowledge? Whose opinion would you weight more heavily?

Moreover, meaningless questions produce meaningless answers. If none thought that Katrina was "the worst natural disaster to strike the United States in their lifetime," would it be any different if all 609 did? Katrina's damage estimate has swelled to $125 billion (the last I've heard), but did people calculate that independently? Or were they merely regurgitating figures, which they may have gotten from the same news that's now polling them? Granted, Katrina's damage is nothing minimal, but mainstream media is telling us that this could be the costliest disaster in a long time while throwing out 12-digit figures. So will a person really reply to a poll, "Gee, I'm not sure it was the worst"? I'd personally put more weight in insurers' initial guesses, not everyday people who happened to be home when CNN called, or who click a "vote" button on the CNN website. Then there's determining the average age of those surveyed, so that we may know how the phrase "in their lifetime" fits in. It necessarily excludes particularly terrible disasters like the hurricane that struck Galveston in 1900, killing over 8000. Certainly people then thought that was quite a terrible disaster, as they did about the San Francisco earthquake and fire in 1906.

It's an old joke in economics that "Talk is cheap because the supply far exceeds the demand," and it applies to polls and surveys. People can give their opinions in polls and surveys without having to carefully consider their answers, or the consequences of giving a wrong answer. They have nothing at stake. Recently in one of my entries on the energy bill, I took note of the difference between what new cars people told surveys they planned to buy, and the cars people actually bought. People can give a quick answer that sounds good, or the sample can be too small and/or too skewed to represent the population with any usable degree of confidence. Either way, the poll results were wrong.

Bryan Caplan last July had interesting thoughts on the reverse situation: teachers getting evaluated by students will cater to a few students at the expense of the rest, because a few students have undue power to destroy the average. The pollsters, in other words, want to influence the results. I myself never encountered that when evaluating professors, though in hindsight there may have been a couple. My dilemma was not having sufficient degrees for my answer (one problem Capital Freedom pointed out), such as a mediocre professor who wasn't as bad as 2 out of 5, yet not quite a 3.

All this proves why prediction markets, covered by my friend Chris Masse, are so much better than polls. For example, they correctly predicted Bush's 2004 victory when many leading polls said otherwise; Chris' site has many other examples. Whether it's a bet on Tradesports or the purchase you actually made versus the idealistic thing you told a pollster, there's a remarkable difference when your own money is on the line: you give more consideration to which decision you make, acquiring as much information as possible to reduce your risk of giving a wrong answer, and unlike anonymous polls, there are definite consequences to your answers. (This is similar to what Milton Friedman has said about government spending the people's money.)

Be careful about pseudo-prediction markets. Don Luskin questioned a need for ConsensusView to track stock markets when they, by definition, already represent a consensus. If you take a look at the ConsensusView site, you'll see it's just a huge poll. Chris' simple-yet-brilliant reply to Don: "Play-money prediction exchanges don't have human market makers, and it's those who have predictive power."

Another problem is assuming the masses are always correct. In highly subjective issues, the results will be skewed because the people's concepts and knowledge are themselves skewed. President Bush's approval rating has fallen to 39%, presumably because "his response" to Katrina is perceived as too slow. How many are those who believe Kanye West's race-baiting slander that "George Bush doesn't like black people" and was intentionally slow because of racism? How many are stupid enough to fall for Ted Kennedy and Harry Reid's partisan inanity?

It doesn't even take that. It only requires a misguided belief, common to most Americans today, in fact a paternalism long-lived for seven decades, that they are "entitled" to help via "government money." There's no such thing, though. The money comes from other taxpayers, and perhaps more people realize that today, which is why it's seemingly less fashionable today to think you "deserve" help from the government, regardless. Ah, but say that you're "entitled" because you pay your taxes, and you won't have to do it in the name of heaven (or Christian charity) to justify in the end that you receive more from government than you pay in taxes.

Remember that only the top half of wage-earners pay federal income taxes. Now consider that the federal government this year will spend about $2.473 trillion (including deficit spending) on 290 million Americans: anyone paying less than $8500 in federal taxes is, therefore, benefiting at the involuntary expense of others. This ties in to prediction markets' superiority in requiring that you put up your own money: so many people who vote for political candidates have little or no tax liability. They have no incentive for the candidate who will most wisely pass good, fiscally sound legislation, but candidates have every incentive to promise the most government spending that others will pay for. Hmm, what was that again that Caplan said about pollsters trying to influence the results?

2 Comments:

Blogger TKC said...

Perry says: "I question how these "609 adults" are so qualified that their opinions are important enough to be recognized in a poll. That's not a personal insult, merely an objective observation that these people may not know much at all about the issue."

This points to the fact that the collective can never speak for the individual. If it turns out that they agree then it is mere coincidence. With a group the size of 609 people you are almost certainly going to get some disagreement on all manner of topics. Does this make the majority of these people correct? No. What applies for one person may not apply for another. It would be insulting to think it would work this way but that is one of the cornerstones of socialism. That the collective knows better what your individual needs are.

Also, leveling something that is damaged beyond repair and building something new in its place is not all that contradictory. The WTCs were damaged beyond repair yet we're determined to rebuild them, or at least something, in their place.

Suffice it to say, IMHO, polls are almost always worthless. I had this to say about an LGF poll: "LGF comes up with its own meaningless poll in response to a Daily Kos meaningless poll. How should one vote in the LGF poll? I think head moonbat Markos is doing a great job... of making the reactionary left look childish, angry, and stupid, which I think is a good thing. Is it good for the Democrats? Obviously not, which should garner an negative vote. But then again, I encouraged the Dems to select Dean for president and then again for chairman because I thought he would screw things up. I'm just a dirty rotten scoundrel that way."

Sunday, September 11, 2005 9:25:00 AM  
Blogger Perry Eidelbus said...

CF's point, though, is that we haven't enough information to judge what the answers really mean. Is New Orleans "beyond repair" in that nothing can be done, or that it can be rebuilt? When the questions themselves are almost open-ended, so the answers will be too.

Monday, September 12, 2005 12:34:00 AM  

Post a Comment

Subscribe to Post Comments [Atom]

Links to this post:

Create a Link

<< Home