Bias: The NINJA
I read something earlier this week about how bias plays a bigger role in your decisions than you know. It brought up a specific example of an experiment done, all with references and good stuff like that. I thought it’d be fun to recreate the experiment. But, you know, without all the proper science stuff. Still, it was fun.
I set up two web based tests and had as many people as I could find try one or the other.
Here is test 1: http://subprocess.net/number_test
Eight out of the nine people I got to try it got it wrong. Only one person got it right. Examples of wrong answers I got:
- Each number has to be +2 the previous number
- All numbers have to be even and in order from low to high
- All numbers have to be + or – 2 of another number in the set, they can be in any order
- All numbers have to be even, order doesn’t matter
The answer? If you haven’t figured it out yet: any three numbers makes a valid set.
So now I was curious… maybe it was just too hard to come to that conclusion? So I made a second test: http://subprocess.net/number_test2. Notice the only difference is the lack of examples of “valid sets”.
I only managed to find 4 people who hadn’t already been exposed to my first test. All 4 of these people got the answer right. Most said “all sets are valid”, which I took to mean “any set of three numbers is valid.”
I thought it was quite interesting. The same exact script with the same exact instructions… but such amazing differences in results. The only difference between the two: the first test pushes a bias on you with the “example valid sets”.
And this is an example of a bias being imposed on you. Think of all the things you might be getting wrong because of a bias you yourself impose on you.
Send the first link to some friends, then send the second link to other friends. See if you find the same thing and let me know.