Mlodinow Cocktail
Posted: May 11th, 2009 | Filed under: Baseball | 33 Comments »
So, you have probably been wondering about the absurd poll questions on the sidebar … what do they mean? Well, to be honest, it was a not especially successful attempt to redo a pretty cool experiment that Dr. Leonard Mlodinow did in his talk the other day. But the results are interesting.
First the experiment: Mlodinow split the room up into two groups. He had Group A look at the screen in front while Group B (which I was in) turned their backs.
He showed Group A two questions.
1. Are there fewer than 180 countries in Africa?
2. How many countries are there in Africa?
Now, I cannot relive the experiment here … but I do want you think about those two questions. Are there fewer than 180 countries? OK, how many do you think there are? Get a number in your mind.
He then had all the Group A people turn around and face the wall while us Group B people got to look at the screen. He asked us two questions. I’ll tell you those in a minute. But before I do, I want you to ask yourself, and be honest: Did you guess more than 29 countries in Africa? If you did … good for you. There are roughly 55 countries in Africa.*
*If you go on the Internet you will get numerous answers to this question, ranging from 48 or so to 60 or so. It’s ever changing.
But our mid-point is 29 countries … because that’s the average number you get from the second group. We were asked these two questions:
1. Are there more than 5 countries in Africa?
2. How many countries are there in Africa?
So, yes, the average from Group B was 29. Interestingly, the average from Group A was roughly 54 ,which is pretty close to correct. But Group B consistently undershoots … and the reason seems to be the way that first question messes with the mind. By asking “Are there more than 5 countries in Africa?” a not-so-subtle suggestion is placed in the mind that there are not as many countries in Africa as you might think. I’m embarrassed to say how low I shot … I guessed 24. I cannot believe I fell for this. But then I saw the guess of the woman next to me … she guessed that there are seven countries in Africa. Seven.
I tried a similar online experiment here … with moderate success. The question built around the batting averages of Ted Williams and Tris Speaker. I picked those two because they are both hall of famers, both prominent (Williams, of course, significantly more prominent) and they had almost exactly the same lifetime average:
Ted Williams: .344
Tris Speaker: .345
Now, first point: The plurality of brilliant readers picked the correct batting average for each player. Forty-five percent of you correctly got Williams’ .344 batting average. Only 24% got Speaker’s, but that’s not surprising — Williams, as mentioned, is a much more prominent player in baseball history. Anyway, 24% is still more than picked any other number.
But I did try the Mlodinow trick — before the Williams question I asked “Did Ted Williams hit higher than .400 for his career.” The wording on that was a little bit off … it would have been easy to read “Did Williams hit higher than .400 IN his career,” which, of course, he did. But it was too late to fix it. I was trying to establish the high number there — like asking if there are fewer than 180 countries in Africa.
Then, before the Speaker question, I asked “Did Tris Speaker hit less than .300 for his career?” I figured that would be enough to get some people to question themselves.
Like I say, it’s an imperfect experiment. I think naturally people would think Williams hit better than Speaker. I also think that the pre-questions were not quite as effective as I would have liked.
Still, there might be a slight effect here:
Average guess for Williams: .349.
Average guess for Speaker: .333.
Kind of interesting. It probably wasn’t worth all that effort.
Loading ...
Your experiment was partially compromised by your audience. Mlodinow’s effect would not show up if you asked the Africa question to geography buffs, because they are too familiar with the subject matter to be misled by any implication in the question. Now you have asked an audience largely composed of sports fans about famous, Hall of Fame baseball players.
A better try would have been to use a couple of the best-known 19th century and/or deadball players – say, Willie Keeler and Hughie Jennings. Casual fans will know these guys went into the HoF, but could have sufficient uncertainty about their specific stats.
Or, a better example: No looking, did Home Run Baker hit more than 40 HR, career? Group 2, did he hit less than 200, career? Then, how many home runs did Home Run Baker hit?
I want to build on Gunthros (#1), but first let me disagree.
I am more than a casual fan and I have no idea (off the top of my head) who Keeler and Jennings are.
Joe, I think that it was a good idea, but the examples were poorly chosen and worded.
If you had asked about career yardage for a couple of great NFL recievers, you have had a stronger showing. Like Keyshawn (10,571) and Andrew Risen (10,205). You’d likely preserve the hint of knowledge — who doesn’t remember these guys, afterall? But who knows their career stats?
Of course there’s there other problem. In the orginal questions, the two nudges were off by orders of (binary) magnitude. One was 4x too great, and the other 10x too small. This are some BIG nudges. Can you really get that kind of nudge for a sports related question? Perhaps career earnings?
The Wisdom of Crowds — an amazing book that covers these phenomena.
It’s amazing how much we’re affected by things like this, and the funniest thing is that if you ask people they’ll say that the previous question (giving them 180 or 5 as a starting number) had nothing to do with their guess, even though it clearly did. We really have no idea what influences our decisions. Scary.
I couldn’t begin to suggest better comparisons (or framing questions, for that matter). But I’m glad you made the effort. The overall idea that the context frames the comparison is often overlooked and certainly worth examing. This is yet another way in which you’ve expanded the way I look at baseball statistics. Thanks!
Personally, I knew the answers to questions 1 and 3 (higher or lower), but not the numbers. I could eliminate the highers ones from what I knew (Cobb’s all-time high career average), but had to guess beyond that. I guessed Williams wrong (.357), but seeing the overwhelming majority picked .344, I figured Speaker’s average was probably around there somewhere.
Home Run Baker would have been good. The problem with Williams was that I actually knew that he hit .344. I haven’t the foggiest idea how I know, but when I read the question, and before I looked at the choices, the number .344 popped into my head.
However, you did get me on Speaker–I guessed .326.
Thankfully, I was not swayed by the questions (this time) and was correct on both.
But you know what? Even more thankful that Mr. Joe Posnanski is back in business.
Joe,
I believe this effect is called anchoring.
http://en.wikipedia.org/wiki/Anchoring
Wow that’s awesome! I love psychology. I’ll tell you straight up that I consciously picked lower for Speaker, because of the question. I second guessed myself about his BA. I was thinking “Wait, if he hit in the .340’s like I think, then why would anyone ask if he hit above .300?” Which made me change my answer, ’cause I figured it just wouldn’t be reasonable to ask such a question on a legit baseball blog, if it were far from it. You made me doubt myself! I’ll have to stop that.
This kind of reminds me of the famous Asch Conformity Experiment experiment which is outlined, along with four other fascinating psychological experiments, on this website: http://www.stumbleupon.com/toolbar/#url=http%2525253A//www.mindpowernews.com/5Psychological.htm
This experiment is so fascinating because, like the one you’re describing, it shows how easily human beings are affected by what other people think of them. The average person would rather conform to something he is positive is wrong than be though stupid by others.
It worked on me (sort of). I got Williams’ average right because I knew it was high but below Cobb’s (famously .367, now it’s .366) and Joe Jackson’s (in the .350 range). .344 was the only number that looked reasonable.
I didn’t really have a sense of Speaker’s average, though. I was pretty sure he was over .300, but the leading question biased me downward to .326.
I think I’m one of the users that skewed the results because I knew Williams and Speaker’s career averages. And I also know that Frank Baker hit 96 home runs in his career, but got his nickname “Home Run” because of three he hit in the 1911 World Series.
Okay, I guess Baker only hit two home runs in the 1911 World Series. I fail.
I’m not even a Royals fan and this current line of whatever the hell it is has me askin’: “So, where’s the damn Banny blog?”
I’m not ashamed to say, I know more about baseball than Africa. I got both of your non-obvious poll questions incorrect, but I knew that there were around 50 countries in Africa. I did a report on it in 3rd grade in Q.U.E.S.T. class. That was 1987. I am a baseball loving asshole from Johnson County. I also saw Bill Pecota pitch an inning. In person.
Also found it interesting that 6% of people voted first that TW had hit over .400, but only 2% picked the .406 option in question 2. Some probably changed their minds when they saw that 90% of the choices were in the 300s. So without any new information about Ted Williams other than seeing the options, 2/3 of the people changed their vote.
While we are throwing out questions to nobody: can anyone explain why the Royals haven’t retired Amos Otis’s number? We need to retire the 26. And we should also retire 29 for both Quiz and Sweeney, though I guess we should wait until Mike retires for that.
What I am amazed about in the experiment is why the people given the number 150 seemed to be able to nail the question and seemingly be unaffected by the first question. My instinct, obviously unsupported by actual evidence in my possession, is that people were affected by the first question. I wonder what you would get if you had a third group of people who were just asked how many countries are in Africa. My instinct is that their average answer would be 40 or so, so that the people shown the higher number would average about 55, and the people shown the lower number would average about 30.
Hey, how are you avatar-havin’ people havin’ avatars? I don’t see where I would get to upload or link to one.
I’m with Willie (#18). What I find interesting is that 6% thought that Teddy hit over .400 but only 2% selected the only option for over .400.
The same with Tris. 18% selected under .300 for his career but only 13% picked the only available option under .300.
The only way I can explain this is that the people who are inconsistant misread the first questions to say “IN” their career instead of “FOR” their career. Still, 4-5% of people making this mistake seems to be an aweful lot.
BTW, I picked 53 countries. Not too shabbby.
The problem with this is that you have one group answering all four questions as opposed to splitting them among two groups. I know it’s with two different players, but there are different levels of knowledge surrounding those two players as well.
I knew the answers to the easier questions, though I’m actually guilty of misreading (or skimming) the second question. I just assumed it asked whether Speaker hit HIGHER than .300 for his career. I had the last question narrowed down to .345 or .333. I figured it could be .345 because it was one higher than Williams’ .344 (which I knew was the right answer), and the whole exercise might have been designed to point out that – as much as Williams is known for being a great hitter – Speaker actually hit for a higher average.
Of course, I talked myself out of that because I know Joe knows how overrated BA is as a stat and went with .333 simply because it seemed to ring a bell as the right answer for some reason. I wasn’t at all confident in that choice, though, and when I looked it up immediately after voting, I felt like I’d actually duped myself as opposed to being tricked by the questions.
As for the number of countries in Africa, my first guess was 55. No idea if that was an actual piece of knowledge kicking around in some hidden brain cache or just a lucky guess.
I knew both of their averages too. I think a better stat would have been RBIs. Career RBI totals don’t seem as prominent.
I had 45 for Africa, a little low… and .315 for poor Tris Speaker, also a little low. Got Teddy Ballgame, though.
This is even more scary (from the wikipedia article on anchoring):
“As a second example, consider an illustration presented by MIT professor Dan Ariely. An audience is first asked to write the last 2 digits of their social security number, and, second, to submit mock bids on items such as wine and chocolate. The half of the audience with higher two-digit numbers would submit bids that were between 60 percent and 120 percent higher than those of the other half, far higher than a chance outcome; the simple act of thinking of the first number strongly influences the second, even though there is no logical connection between them.”
In other words, ANY number that gets put in your head will influence a subsequent guess or bid. It’s a wonder we ever make good decisions, given how messed up the wiring of our brains is.
I think your experiment worked on me, Joe, because while I knew, as a Red Sox fan, that Williams hit .344 for his career, I wasn’t as sure about Speaker (through I should have been), and I low-balled him at .333.
Hey, at least I nailed the Africa question. Not only was my first thought “somewhere between 50 and 60″, but I consciously left it loose because I know that number shifts regularly, as you alluded to. I guess my poli sci degree finally came in handy. First time for everything.
I forgot Gabon and Ghana, but somehow remembered Sao Tome & Principe. I called Burkina Faso “Upper Volta”
The only island nation I included was Madagascar…
Seychelles, Comoros, and Mauritius count as “Indian Ocean” to me. And Cape Verde as “Atlantic Ocean.”
i guessed 45
I guessed .326 for Speaker, so you got me. I think that so many people got it right speaks to the high level of baseball knowledge amongst readers of this blog.
Check out the book “Influence” by Robert Cialdini. He discusses this psychological effect. It is the same idea jumping in a pool after sitting in the hot tub…it will feel much colder than it really is.
(This is, of course, a standard hotel staying experience)
In other words, ANY number that gets put in your head will influence a subsequent guess or bid.
“My client, Stephen Strasburg, will require a $50 million dollar deal to sign!”
Nailed Williams, low on Speaker. Although I did guess 24 countries for Africa (same as Joe). I think that I was influenced by reading this blog too much more than the 1st question.
I’m a bridge player, so I remember numbers. I was right on both Speaker and Williams. On the other hand, if you ask me to remember names, or especially faces, then I’m in trouble.