Nov 17, 2011

On Human Error

Go here to read Michael Lewis' (of Moneyball fame) great Vanity Fair article on Nobel Prize-winning psychologist Daniel Kahneman.  It's utterly brilliant.  The power and often failure of human intuition and reasoning. Good stuff.

Here's an excerpt:
In 1983—to take just one of dozens of examples—they had created a brief description of an imaginary character they named “Linda.” “Linda is thirty-one years old, single, outspoken, and very bright,” they wrote. “She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in antinuclear demonstrations.” Then they went around asking people the same question:
Which alternative is more probable?
(1) Linda is a bank teller.
(2) Linda is a bank teller and is active in the feminist movement.
If you answered (2), aren't alone. In a poll conducted, 85% of respondents agreed with you.  Once you really think about it, you will realize why that isn't the correct answer.  Or just read the article to get smarterer.

See? It ain't only sandwiches and baseball here. 

PS - But if you NEED the sandwich stuff, here it is. You are welcome.


Slyde said...

i cant read this article since im still staring at that colon picture...

B.E. Earl said...

Slyde - You love your colons, that is true.

sybil law said...

I guessed 1. Me so smaaart!!

B.E. Earl said...

Sybil - But did you guess, or did you KNOW it was the correct answer? ;)

Candy's daily Dandy said...

I read ahead too quickly and knew two was the wrong answer before I had a chance to guess.

I hate it when that happens.

RW said...

I've been trying to figure out what I'm being asked to do here for two days now.

Mrs. Hall said...

I don't get it.

'something's going on and i'll prolly never get it'

10 points for that song title.

LegalMist said...

Interesting article, but I disagree with their premise regarding the correct answer to the question they asked (assuming, of course, that the author of the article didn't misstate the question used in the survey).

As written, the question implies that she is either a bank teller *and IS NOT a feminist* or is a bank teller *and IS a feminist."

In other words, given the two options as answers to the question asked, it is clearly true that she is a bank teller (thus their conclusion that (1) is "more probable,") and I'm sure everyone who read the question understood that. But the normal human interpretation of the sentence is that number (1) implies that she is also NOT a feminist. And if this were a conversation between two people, instead of a survey in which economists are trying to trick people, that interpretation would likely be correct!

Now, whether it's a good idea to assume things not stated is a different issue, but the fact remains that most (apparently about 85%) people apparently interpret the question to mean that it is clearly true that she is a bank teller, since that is part of both answers. (And probably leaves most folks wondering why it had to be stated, when that part obviously has to be true).

Thus, for most normal people, the real question being asked is, is it more probable that she is, or is not, a feminist, in addition to being a bank teller?

Based on the description given of her, I think most people correctly deduce that it is more likely that, in addition to being a bank teller, she is also a feminist.

That is a different question than the authors seem to be asking, which is whether it is more likely that every part of sentence (1) is true, or that every part of sentence (2) is true. I bet if they asked the question that way, they would get a pretty near 100% correct response rate that it is more probable that every part of sentence (1) is correct -- for the reason stated -- that it is impossible for (2) to be true without (1) also being true.

But that is just not how we normal people communicate. (Although I will leave open the possibility that it's the way economists communicate, and thus why 85% of the population doesn't understand them.)

B.E. Earl said...

Candy - cheater!

RW - Not much, really. Just an article I found interesting.

Holly - I lose. No clue.

LM - Yeah, that's what it's all about. "Normal" human comprehension, and why it is sometimes incorrect. Answer 1) doesn't imply anything. It's you the reader who infers that it might be stating that she is NOT a feminist, but all it is saying is that she is a bank teller. Nothing more.

LegalMist said...

But there's a difference between "normal" / possibly incorrect human comprehension of events, probabilities, data, or scientific principles and "normal" / generally correct human comprehension of the language we speak.

Just because language *CAN* be interpreted ridiculously narrowly doesn't mean that it is "correct" to do so. That is why we recognize things like "lies of omission." If your wife asks you, "where were you last night" and you say, "Oh, I had to work late," when in fact you did work late and then went to your girlfriend's house for 6 hours, you have technically told the truth, but I am willing to bet that at least 85% of people hearing that response (including the wife) would interpret it to mean that you *only* worked late and did nothing else (and would *correctly* accuse you of a lie of omission). [Not that you would do such a thing, of course!! :)]

Obviously, that isn't precisely what was stated, but was clearly the understanding that the person intended to communicate -- and probably successfully did communicate.

People may make mistakes in communication, including reading things into statements that aren't there. But I submit it would be *more* erroneous, in normal human communication, *not* to interpret the "I had to work late" statement (or the survey question) the way most people (85% of survey takers, apparently) interpret it. The whole point of language is to communicate, and it is as much the speaker's responsibility as the listener's, to make sure that meaning is properly communicated.

After all, if it is true that 85% of people understand the survey question to mean "is it more probable that she *is* a feminist in addition to being a bank teller, or that she *is not* a feminist in addition to being a bank teller?" then if you ask that question as written, you have to expect that 85% of your listeners will understand that you are asking "is she or isn't she likely to be a feminist in addition to being a bank teller?"

So if you ask that question and you actually mean something else, such as "statistically speaking, for which sentence is it more likely that every portion of it is true?" -- aren't you the one making the communication error? I.e., isn't the survey question faulty, because it fails to measure what you want it to measure?

And if you want to ask the more precise question "for which sentence is it more likely that every portion of it is true?" then you need to ask the question that way so others will understand the question. Then, if you still get 85% of people answering that (2) is more probable, you will have proven that they misunderstood the facts and probabilities -- rather than that they applied the usual rules of interpreting English languag and that the survey question was faulty.

In other words, I don't think people are misunderstanding the actual facts -- that, given the two options stated, it is 100% likely she is a bank teller, and somewhat less likely that she is both a teller and a feminist (and even *less* likely that she is both a teller and not a feminist). I think they are simply interpreting the question according to the normal underlying rules governing communication in English, that most of us have learned to apply when speaking to others. And under those rules statement (1) absolutely DOES imply that the speaker / asker is stating she is NOT a feminist.

Except that economists who put together surveys apparently haven't learned those rules.

Lawyers are guilty of that nit-picking and failure to properly interpret, as well. We are famous for noticing the warnings that weren't given, and bringing lawsuits based on it, when most of the normal people in the world would think, "of course I shouldn't use my hair dryer in the shower."

It doesn't make it "right." :D