Wednesday, June 15, 2011

Responsible Democracy

South Africa has just had a local election and the US is already starting to gear up for the 2012 Presidential election. Almost everyone who has something to say about it will tell it’s very important to vote. About 80% of the US population considers it a very important obligation for a citizen to vote but here’s the thing – of those who consider it so import only 57% think it’s also very important to be informed about matters relevant to your vote. It’s no surprise then that voters are quite astonishingly ignorant.

Some examples of the degree of ignorance – in 2000 26% thought Al Gore was a Conservative and 23% thought George Bush was a Liberal; 40% and 60% respectively don’t know that defense spending and social security are the top two budget items; only 50% knew which Party controlled the Senate before the election; barely 35% could identify British Prime Minister as the post held by Tony Blair, and 58% knew little or nothing of the Patriot Act. Some 25-29% of US voters are literally ‘know nothings’ in that they did no better than pure guessing on a 31 item political questionnaire featuring items like those above. If some easy questions, mainly personal information on candidates, are dropped the proportion of ‘know nothings climbs to 35%. Even this overestimates their knowledge because the remaining questions were still fairly basic. For example social security is virtually never mentioned as a racial issue. Yet because blacks have a lower life expectancy, and they pay the same social security rates, there is a substantial regressive redistribution of income from black workers to white retirees. If this sort of knowledge and insight were included in political knowledge assessments the level of ignorance would be shown to be far worse.

How can one judge whether the candidate you vote for is good for you, or any one else, if you don’t even know who the conservative and liberals are in the election, or anything about their programs? For all you know the other guy’s program might be much better for what you value. How would you know which candidates, if any, aim to deal with anything really important if you don’t actually know enough to say what is or isn’t import? Mostly people use shortcuts, like relying on Party affiliation or opinion leaders and activists. Unfortunately it’s difficult to gain much information about Party policy effectiveness from experience with a few governments, and opinion leaders tend to have more extreme opinions than the general voter and have an incentive to exaggerate the importance of their pet issues. Sometimes people vote on the basis of the state of the economy but many are even ignorant about that. For example about 60% couldn’t tell you whether unemployment is improving or worsening. Even if they knew the state of the economy they wouldn’t be able to say whether the state had anything to do with the previous government’s policies, and even if they knew that they would still need to know if the other Party wouldn’t do better.

Some commentators have claimed this widespread ignorance isn’t a problem because the ignorant voters constitute random noise and the votes of the informed few therefore make the overall vote correct. This assumes that the votes of the ignorant are truly random i.e. make errors equally in all directions and that the informed voters represent the interests of the general electorate. The evidence is strongly against either assumption being true. The very shortcuts voters use creates systematic biases e.g. favoring known politicians over new ones or loud special interest activists over good sense. Informed voters are decidedly not representative - they are very slanted in terms of gender, race, income, education, ideology and age, and of course lobbyists with special interests tend to be very knowledgeable politically.

There are also many issues about the rationality of voter ignorance or of voting itself and various paradoxes produced by voting systems, but I won’t discuss these now. Suffice it to say that a lot of ignorant voters aren’t just random noise which ballot distills the good sense of informed voters. Ignorant voters tend to be systematically biased away from their own optimum positions i.e. support bad policies and programs. To the extent that politicians give them what they want they therefore cause real harm. Voting is far from being an important obligation if you are ignorant. If you don’t know the facts or understand the issues and how government works, voting is downright irresponsible.

OK so how much irresponsibility is there going around? Look at the graph below.



The pink line shows how well informed voters think they are by IQ (the self judgments are weighted by some knowledge testing). Like knowledge and understanding in every field political knowledge increases with IQ so we should be grateful that the probability of voting (the dark blue line) also increases with IQ. Indeed the yellow line shows that the probability of being politically savvy if you vote rises with IQ. The purple line curving downwards shows the probability of thoroughly ignorant people taking the trouble to vote. The peak probability of this curve is in the same IQ range where crime is most probable - apt for choosing to be irresponsible. Finally the light blue line is the probability of voting if you are well informed plus of not voting if you are ignorant – hence the label, ‘responsible’. Note that as many as 1 in 5 of even the very brightest among us is politically ignorant, and is prepared to vote irresponsibly. Happily responsible voting never drops below 50% at any IQ. The overall figure for responsible voting behavior is about 57%. So just more than 2 in 5 people know a lot and fail to make that knowledge count, or they know little to nothing and inflict their ignorance on us politically.

Responsible voting increases with intelligence so it would be informative to take the policy preferences and votes of the most intelligent within every demographic group of interest and then weight those votes to reflect the relative size of each group. That way we could estimate the intelligent, informed and least biased vote in a way that represents everyone’s interests.

Let me show you another exercise where the intelligent vote tells us something interesting. Look at the next graph. Each dot is the ratio of the proportion of voters with IQs above 120 voting for a Democrat or Republican presidential candidate, over the proportion of the total vote that went to the same candidate. This is for whites only because blacks tend to vote exclusively Democrat.



Don’t make too much of the fact that the Democrat line is mostly above the Republican line. The truth is that if we looked at a stupidity ratio instead we would see the same thing because Republican voters tend to cluster around moderate IQs and Democrat voters are disproportionally either very bright or very dull.

What I want to show is that the ratio at any point predicts the Party of the next president quite accurately. The rule “if the Republican line gets very close to or rises above the Democrat line for any election it will win the next presidential election – other wise the Democrats will win” gets 5 out of 7 correct. If I add a rule saying that “if the previous election had the Republican line at least close to the Democrat line then the Republicans would win the next presidential election regardless of how much the Democrat line exceeds the Republican line in this election” then the hit rate is 7/7. Victor Serebiakov (former Mensa International Chairman) said something similar in a report on the voting of Mensa members in both the UK and the US (going back further than I did). He showed that the Mensa vote tended to anticipate changes in the general population vote very well. Based on my two rules it looks like the Democrats will win the next US presidential election. Currently election futures markets are saying the same thing.

1 comment: