Friday, January 27, 2012

representativeness bias

Many of our judgments involve classifying or categorizing individual persons or things. The representativeness bias manifests itself when we take a few traits or characteristics of someone or something and fit them to a stereotype or model. For example, when told only that a man is quiet, shy, reserved, and self-effacing, and forced to guess whether he is a salesman or brain surgeon most people would probably choose brain surgeon because their stereotype of a salesman is of an outgoing, gregarious person. But the odds of any given man being a salesman are much higher than the odds of his being a brain surgeon, so the probability is greater that the fellow is a salesman. This example is representative of the examples of representativeness presented by Tversky and Khaneman. In real life, a snap judgment about a person's occupation based on knowing about that person only a few personality traits would be hasty and unwarranted. A reflective judgment would require at least some assessment of the accuracy of the reported traits (how reliable is the source of information?) and the base rate for salesmen and brain surgeons. Most people would not know what percentage of the population are salesmen and what percentage are brain surgeons, but most people would know that the base rate for salesmen is much higher than that for brain surgeons.

If told only that a woman is quiet, shy, reserved, and self-effacing and forced to guess whether she is a sales clerk or a brain surgeon, I think many people would guess that she's a sales clerk because their stereotype of a brain surgeon is of a male and their stereotype of a sales clerk is of a female. The guess would be right most of the time, but for the wrong reason.

When I was in graduate school in the late 1960s at the University of California at San Diego, the philosophy department hired its first black teacher. I arrived there about the time Angela Davis was finishing up her work with Marxism scholar Herbert Marcuse and the daily news was filled with stories about civil rights and anti-war protests. Students were coming from all over the world to study political philosophy with Marcuse and Stanley Moore. I remember how most of the graduate students assumed the new black teacher would be coming to teach political philosophy and would be radical like Bobby Seale or Eldridge Cleaver. He didn't fit the model at all. I don't remember his name, but I remember he came from Ohio State and his interest was in analytic philosophy, much to our shock and dismay. He didn't fit our stereotype at all.

The key to avoiding the representativeness bias is to be open to the possibility that the case before you isn't typical. Force yourself to consider other possibilities. Jerome Groopman, M.D., gives the example of a doctor who failed to diagnose a cardiac problem with a patient because the patient did not fit the model of a person likely to have a heart attack. The patient complained of all the things a person with angina would complain of, but he was the picture of health. He was in his forties, fit, trim, athletic, worked outdoors, didn't smoke, and had no family history of heart attack, stroke, or diabetes. The doctor wrote off the chest pains the patient complained of as due to overexertion. The next day the patient had a heart attack.

We should remember the representativeness bias when we read things like the following:

Atheists are one of the most disliked groups in America. Only 45 percent of Americans say they would vote for a qualified atheist presidential candidate, and atheists are rated as the least desirable group for a potential son-in-law or daughter-in-law to belong to.
When people are polled about "atheists" (or "Christians," "Jews," or "Muslims") the concept that comes to mind for those words will differ depending on what stereotype is at work. Many people who say they would not vote for an atheist might actually vote for an atheist. Why? Because the atheist that runs for president will be a real human being, not a stereotype. There will, of course, be many people who cannot overcome their prejudices against atheists, gays, Christians, Jews, Muslims, Mormons, adulterers, etc., and no amount of empirical data or experience will change their minds. But not everyone who thinks of a stereotype when asked a question about a group by a pollster is hidebound and bigoted. Some will be open to changing their minds as they find out more about an individual who is an atheist, gay, Christian, Mormon, adulterer, etc.

Will Gervais of the University of British Columbia, the one whose study found that 45% of Americans say they wouldn't vote for an atheist running for president, thinks atheists are not trusted as much as theists. I would temper this comment with a reminder that this opinion applies to stereotypes not to real individuals. If a corrupt Christian were running against an upright atheist, I suspect that many people who trust "theists" more than "atheists" would vote for the atheist. But many people can't be trusted when they answer poll questions and it would not be surprising to find that many religious people who say they trust  African-Americans more than adulterers would vote for the adulterer rather than the African-American in certain circumstances.


  1. This is related to what I consider to be the most frustrating of modern day logical quirks: Pigeon-holing, and labeling.

    You can't really begin to dismiss something - a cause, movement or idea - until you've properly given it a dismissive label. While I'm not personally, say, a "denier", "truther" or "tea-bagger", I refrain from using those labels because it seems petty and unimginative, and I hate it when it comes back to me, such as when I express my thoughts about the wantoness of war and become a "'cut and run'er".

    We seem to just follow the leads of TV personalities who make a living enraging and polarizing the populace.

    Regarding bias: I'm the CEO of an IT services firm. Last year I traded in my Acura for a Chevy pick-up, for several reasons that make a lot of sense to me and don't need to make sense to anyone else. I almost wish I hadn't - it's been a year of almost ruthless BS, mocking and disbelief from my so-called peers.

    Thanks for the blog and the newsletter.

  2. Stereotyping and labeling simplifies communication but it distorts reality. Sometimes, as in the case of biased TV personalities paid to enrage the populace, the labeling is intentional. Even owning a pick-up truck invites stereotyping, as does driving a Prius. Whattayagonnado?

  3. What about for non-human entities, such as "pesticides"?

    If one excoriates "pesticides" by citing the horrors of DDT (insecticide) or "Agent Orange" (herbicide), for example, is that an instance of representativeness, by having the bad case function as the stand-in for all pesticides?