Mar 25, 2011

Science: How To Fake It

MAD ART LAB: “So you want to publish a fake science paper. Of course you do. Who doesn’t? But how do you go about it? Well, it’s a lot easier than you think. Just follow these simple steps…

Step 1: Pick a Subject
This is important. You’ll need to choose something that’s both popular and wrong. Things like “sticking needles in your skin cures disease” or “pets can telepathically detect when their owners are coming home” are pretty good. For this tutorial, though, we’ll use “common objects can bring you good luck” as an example.

Step 2: Prepare Your Experiment
Your experimental setup, of course, depends on the subject you’ve chosen. For this one, we’ll need a pair of dice and a bunch of objects to serve as good luck charms. The more the better. We want to run lots of trials, so gather up everything you can find: four leaf clovers, horse shoes, that pebble you found last week, Gerald your pet hamster, whatever happens to be lying around.”

“Step 3: Run Your Trials
Here’s where the magic starts. What you want to do is run lots and lots of trials, each with a small sample size. This will increase your chances of a false positive.

Step 4: Hunting The Wily Anomaly
Taken as a whole, your data will look pretty mundane. The chances of rolling, say, double sixes are 1 in 36, so you’re going to get about 2 or 3 per 100 rolls, give or take. Your results probably fit fairly well onto a bell curve, with most of the data points clustered in the middle and a few outliers on either side. But wait! If you rename some of those outliers “anomalies”, you’ve suddenly got a phenomenon!”

Continue reading the rest over at Mad Art Lab

No comments:

Post a Comment