Friday, November 10, 2006

the myth of the rational voter

This is a great article on its own merits, but also has a lot of Human Factors to it, especially considering that it was written by an economist.

If political bias were systematic, then politics would work. The average voter would be in the middle. But Caplan found that voters make the same mistakes, in some cases way off the middle. They significantly overestimate the amount spent on foreign aid and underestimate the amount spent on entitlements. I wonder if this is confirmation bias (their ideology wants to believe one thing and so they only focus on evidence in support of that - such as listening to partisan talk radio) or if the media misleads people by selecting stories based on salience rather than providing a full source of information.

Another finding is that expert economists have very different views than voters. They tend to be more libertarian. Caplan says this shows voters are wrong because economists are probably right. But I suspect that they are also wrong to some extent, because they trust theories that do not always hold up in the complex, naturalistic world in which we live. I am a free market guy, but I think most economists put a little too much faith in Adam Smith's invisible hand.

Interestingly, he says that voters tend to vote based on what they think is best for society, rather than what is best for themselves. This is encouraging in terms of belief in people's inherent goodness. But if most voters have no clue what is really best for society, it doesn't really help.

For example, voters are susceptible to salience bias. A great story on the news can overcome dozens of small experiences in the real world, especially when focus of the story is on the policy issue but in the real world we are just trying to get on with our lives.

This can also manifest in the effect magnitude. If 100 people lose their jobs but 30 million people save $10, it is good for the economy, but whose story is more salient on the news, the laid off retail worker or the family who saved a few bucks at Walmart?

One conclusion we can draw from all of this goes against what I have always believed (wow - Caplan changed my world view to some extent). I always thought that people should become familiar with the issues and vote for the candidate who supports those policies. But in fact, few voters really have the time to do this well. So really, we should select candidates based on their core values and trust that they will study the issues for us and make the best decisions. Then we shouldn't question them on it unless something seems really fishy.

A different Human Factors issue he talks about is that voters tend to be very confident in their judgments, even getting angry if you contradict them, when they clearly have very little evidence to support their views. He talks about how it is much easier to believe in the emotionally appealing side than the correct one (which requires studying). He uses as an example the immigration debate, where it is easier to blame "sneaky foreigners" than the laid off worker.

Also, there is very little feedback when you are wrong. When do we ever get to see a controlled study where two otherwise identical countries try different immigration policies and then analyze which one worked better? Econometricians do this, but most people don't read the results and wouldn't understand the statistics even if they did read them.

To prevent this problem, Caplan suggests putting more power in the hands of unelected officials who are experts in their field, such as the Fed and Supreme Court. This is OK as long as you are consistent. I suspect that the libertarian Caplan would not want nutritional experts deciding to limit his ability to eat trans-fat laden foods at McDonalds, even if they know better than he does that the negative health consequences outweigh the great taste and that these costs transfer to all of society in the aggregate.

Maybe we need to replace all political consultants with Human Factors practitioners. The pay would be better. At least, that is my inexpert perception :-D.