Sunday, December 6, 2009

Flimflam innoculation 4: Proof by example

If you're still not convinced that the AssBlaster can shave pounds off your hips in just 5 minutes, listen to what these people have to say.

Person A: "I was amazed at how quickly my weight dropped."

Person B: "It worked like a charm for me... you gotta try it!"

The part of the interview they don't show you...

Person A: "Come to think of it, that was around the time of my chemotherapy."

Person B: "The AssBlaster was a great warm-up for my Ironman triathlon training."

The point is, testimonials do not give you all the information you need to establish if the claims are true. Instead, testimonials are directed at the standard loopholes in your psychology. They are not meant to prove anything, but are meant to make you BELIEVE. That's why the skeptical hairs on the back of your neck should stand up when someone starts feeding you success story after success story. Delivering testimonials is often a sign that real evidence does not exist.

If they had a product that really worked, then they would do a blinded and controlled study. Here's a recipe:
  1. First, come up with a hypothesis: eg. "Using the AssBlaster for 1 week will reduce your weight by 10 pounds and your waist measurement by 1 inch".
  2. Then, randomly choose a bunch of people and split them into 2 groups (A and B). Have group A use the AssBlaster, and group B do the same exercise as group A but without using the AssBlaster.
  3. Measure the relevant quantities before and after a trial period (eg. weight, and waist measurement). The people doing the measuring should be "blinded"... they shouldn't know who is in group A and who is in group B.
  4. At the end of the trial period, study the data in a quantitative, statistical manner to see how likely it is that the data supports your hypothesis. You're look for a bigger drop in weight for members of group A compared to members of group B. Likewise for waist measurement.
The presence of a "control group" (B, in this case) is very important. It allows the scientists to establish an association between the hypothesized cause and effect while minimizing other confounding factors. This is really the only way to test the claim, unless you know some basic mechanism that is so plainly obvious that everyone believes you already (eg. "Not drinking or eating for a week will reduce your waist size by 1 inch.")

(For more examples of testimonials, see my post on the Canadian Decompression and Pain Centres.)

In summary: If some claim sounds too good to be true, you might be right. If the only evidence they offer in support of their claim is testimonial, run... FAST!


No comments:

Post a Comment