In the previous post I briefly introduced the concept of The Smart Vote. In this post I spell it out in detail because I will be using it frequently in future posts.
There are always arguments about political policy, or about social, economic and religious issues, with people on all sides thinking their own position is the one that makes obvious sense; but can one objectively establish if one position make more sense than the others, and if so can we identify it? I think differences in intelligence can help point the way.
The essence of the Smart Vote concept is as follows.
If there is no difference of opinion by intelligence then reason is not relevant in deciding between them and none of the opinions being considered is more correct than any of the others. However if opinions do differ systematically with intelligence then relatively more correct or better alternatives probably do exist, and that they are those relatively more favoured by the more intelligent. Statistical differences in the independent opinions of people of different intellectual ability point to the most reasonable responses to controversies.
Intelligence as correctness
An essential attribute of any concept of intelligence is that the smart should be better than the dull at finding objectively correct answers to problems, where such answers exist. In fact this attribute is almost always used when attempting to measure differences in intelligence. Tests usually present problems that have objectively correct answers and then require the person to find them if they can. Those who are more adept at this are deemed relatively smarter than those who fail. This inference is only valid when all those taking the test are roughly equally familiar with the kinds of problems used, but in general that condition is largely satisfied.
One well established finding is that the ability to find correct answers to one kind of problem type e.g. arithmetic, says a lot about the ability to find correct answers in other problem types e.g. reading comprehension. Smartness with one problem type tends to go with smartness in all other types – no exception has ever been identified. Being smart is, at the very least, about the ability to find correct answers to all problem types. Even if that wasn’t the case individual intelligence tests (IQ test) present the person with a wide array of problems, so it is only possible to achieve a high IQ score if one is reliably adept at finding correct answers to a wide range of problem types. Getting all the arithmetic and none of the other correct answers will not get you a high IQ score. Likewise it is only possible to get a low IQ score by consistently failing to find most of the correct answers on all the problem types. IQ therefore is a measure of one’s general propensity to find correct answers.
This is all very well when dealing with unambiguous problems that do have clear objectively correct answers, but in most of life the answers are highly uncertain. Is intelligence of any use there? Yes indeed. Somewhere in between objective and uncertain problems there is a kind of item made up of mutilated pictures of well known objects. A mutilated picture is almost like an inkblot test, where what you see is determined by your own psychological makeup. With inkblots you impose your biases onto the picture and sort of see what you want, or need, to see. In theory what you see says nothing about the picture and everything about you. Mutilated pictures however have proved to be superb intelligence test items. People who score high on objective tests are better at guessing the actual original object from random partial bits of its image i.e. the smart are more likely than the dull to impose or project the correct meaning onto the bits and pieces left over.
Similarly, there are an infinite number of correct responses to any number series. Some of course involve less complicated principles than others and it is usually the simplest principle that the test designers have in mind when constructing a number series. Those who are good at objective problems tend to select the correct answers that the test designer had in mind rather than any of the infinitely many others. So again people smarter on problems with a single objective answer are better at finding the best answer when there are many correct answers.
Then there is achievement in life. All things considered, on average (for people who start at the same level) those who end up better off – wealthier, healthier and more respected - will tend to have picked the better alternative whenever they faced a choice. Many of those choices would have been uncertain. The choices may have involved more than one reasonable option or have been controversial. On average, those who are good at finding objective answers are better at these messy choices and do better in life. For example the average of the upper, middle, working and lower classes in the US are at the 69th, 60th, 43rd and 27th percentiles of intelligence respectively. Professionals like doctors, lawyers, engineers, accountants, CEOs and even US presidents are on average at the 93rd percentile. Top thinkers such as elite scientists or supreme/appeal court judges are at the 99.9th percentile.
Men who make choices that see them serving time in jail typically score at the 27th percentile. A large part of that is because dull men are more likely to commit crimes serious enough to warrant jail time. However brighter men are also more likely to make choices less likely to see them get further into trouble at every possible step between committing a crime and going to jail. They are less likely to be identified but if identified less likely to be arrested; if arrested less likely to be prosecuted; if prosecuted less likely to be convicted and if convicted less likely to be jailed. Brighter criminals at large are less likely to be caught and if caught it tends to take longer. If jailed brighter criminals are more likely to escape or be a model inmate and get time off.
Women who give birth out of wedlock average at the 35th percentile of intelligence.
Humour is to a large extent subjective, and whether someone will laugh at one’s jokes or wit is highly uncertain. Nonetheless some people get it right far more often than others and those that do get it right tend to be disproportionately smart. IQ testing of the top 50 or so stand up comedians in Britain found that they averaged at the 97th percentile and none were below the 90th percentile of intelligence.
So, intelligence is relevant and important when it comes to finding good answers to problems, even when the situation is uncertain and the answers fuzzy. If there is a correct answer to a problem then the opinion of the smart person is more likely to be correct than that of a dull person.
Using intelligence to point out the correct answer
Suppose you can’t use the content of the problem to identify the correct solution. This usually means you personally can’t answer it. You can ask for help. Suppose no one is around to ask but you do have data on the answers of the smartest and dullest 20%. Consider the following actual 5 item science test from the General Social Survey. Each question had 4 different possible answers - A, B, C or D. The numbers are the percentage of each row that chose that answer.
The correct answers were 1 D, 2 A, 3 D, 4 A and 5 D.
The test was tough. It was unusual for any item to be picked by more than half of either group and the mean score of the smart and dull groups was 42% and 15%, respectively. Pure guessing is 25%.
Note that it wouldn’t be much help to go on the opinion of individual smart people. On this test even they were wrong more often than they were right. On the other hand the average of the independent judgements of a diverse group of people is often very accurate (Wisdom of Crowds) and betting markets are astonishingly good predictors. Not all crowds are equally accurate however. Quite frankly the ‘wisdom’ of a crowd of dull people can be very unreliable. For example if we went along with the most popular choices of the dull group on each of the test items we would have netted no correct answers at all. On the other hand if we went along with the most popular choices of the smart group we would have netted 4 correct answers out of 5 – much more than the average total score of that group. So the difference between the average ‘wisdom’ of groups of dull and smart people is even starker than the differences between individual dull and smart people. I’ve shown this with a concrete example but it is true mathematically.
Members of each group share a particular level of intelligence (or propensity to find correct answers) and whatever factors they don’t share will tend to cancel out in the average opinion. The difference in the percentage of each group that picked an answer will therefore reflect a distilled difference in ‘propensity to be right’. The Smart Vote makes use of this difference in distilled propensity to find correct answers.
In the Smart Vote it is not the most popular answer of the smart group that matters but the largest difference of opinion between the smart and dull groups. The percentage of smart and dull groups choosing each answer is compared and the largest ratio of the smart to dull percentages is the Smart Vote. For example for Item 1 the ratios were
A 2.4/17.7 = 0.14
B 13/42.3 = 0.31
C 21.8/28.5 = 0.76
D 62.8/14.5 = 4.33
The largest ratio was for alternative D so the Smart Vote for Item 1 would be D – which is the correct answer. The Smart Votes for the other four Items were 2 A, 3 D, 4 A and 5 D – all correct - so the method proved 100% accurate even though the average total score of the smart individuals was only 42%. The probability of that happening by chance is less than 1 in 1000.
I have repeated the above experiment on many different tests with objective answers, always with 100% accuracy. To some extent it’s to be expected that the correct answer would be picked more often by the smart because the item would be a bad measure of intelligence if it wasn’t. On the other hand there is no reason why one of the several wrong answers shouldn’t show a larger ratio in favour of the bright than the correct answer. So there is no necessary reason why the method should be so accurate, but it is.
Going beyond questions with objective answers
Guessing the correct answers to IQ or science test items is not the point of the method. The aim of the method is to shed light on controversies. The answers to test questions are in a sense controversial in that even the very bright clearly don’t initially agree, but they do agree following some explanation. Unfortunately agreement on religious, social and economic controversies is harder to come by.
One’s grasp of the meaning of a controversy and understanding of what is at stake should increase with increase with intelligence, both because of the greater capacity to reason through the complexities and the larger stock of information that intelligent people tend to collect. If differences in intelligence make a difference to the understanding of controversial issues they should also make a difference to opinions formed. At the very least careful intelligent thought should be relevant.
To the extent that ‘correctness’ is relevant to a particular question those with a higher propensity to find correct answers should reach different conclusions to those less able. In other words, if ‘correctness’ is meaningful for an issue then on average the smart and the dull will tend to have different opinions. Logically, no association between IQ and opinion proves that cognitive factors aren’t relevant to the controversy. However the presence of an IQ-opinion association does not prove that intelligence is relevant – it only suggests that it is quite likely. In my view the likelihood is very high, especially in view of how well the method performs on objective questions.
An objection to the Smart Vote
People tend to vote for what is best for them personally and smart people may simply have interests in common that are different to the interests shared by dull people. The defining features differentiating the two groups may not be IQ alone but some interest too, and the average opinion of each group will distil the interest differences as much as they do intellectual differences. For example, IQ related differences in musical taste may simply reflect the fact that IQ relates to social class, and that it is class prejudice that explains the differences musical taste rather than any application of thinking to the relative merits of musical forms per se; or a difference of opinion on welfare measures may simply reflect that dull people are more likely to be recipients, and smart people contributors, to welfare.
On the face of it this is a compelling objection. My first response is that this isn’t always a factor. Political research shows that people frequently don’t vote their narrow selfish interests e.g. the elderly are less likely to vote for social security than the young and women are less likely to support abortion on demand than men. However there are enough cases where narrow interests obviously do play a role for it to be taken seriously. Fortunately this possibility can be dealt with by controlling for interest differences. For example we could control for class when looking at musical taste, income when looking at welfare, age when looking at social security policy, race when looking at affirmative action, etc.
I should point out that though it is possible that selfish interests may produce spurious associations between IQ and opinions they could just as easily hide real associations. For that reason too possible interests should be controlled.
When IQ is correlated to differences of opinion there is reason to believe that some opinions are in some degree better than others and that the most reasonable opinion is the one with the largest ratio of smart to dull favouring it. This is the Smart Vote. A related concept is the Stupid Vote which is simply the opinion with the lowest ratio of smart to dull favouring it. The Smart Vote is not necessarily the opinion most favoured by the smart nor is the Stupid Vote necessarily the opinion most favoured by the dull – it’s the ratio that matters, not the level of support.