Self-Help and Self Experimentation

by F.

There is a fairly obvious problem with self-help books and schemes, even if they are scientifically based: you don’t know whether they are working for you—and if they are working, you don’t know the effect size. So, really, you should keep data on yourself and do some basic statistical analysis.

Back in the day, only the pocket-protector crowd knew how to do this. Fortunately, the tools to do this are becoming more prevalent. For instance, there is StatCrunch, which costs a whopping $8 a year to use. Maybe there are others, too. The Shangri-La Diet forums are the closest thing I’ve seen to this sort of thing, where people post their data sets.

But it seems there should more—a site that allows you to put in your data and computes r and so on. Were I to write a self-help book, I would have a companion site that would:

  1. Allow people to post data;
  2. Aggregate the data so I could study it;
  3. Run basic tests on the data; and
  4. Spit out easy-to-understand translations of what the statistics meant so the customer could use it.

Like retirement calculators—you put in your data and it tells you whether you will hit your saving target. Again, the Shangri-La Diet forms are close, but I’m thinking it could be more Web 2.0-ish.

How many times do you have conversation with someone who is trying some self-help remedy and yet is not tracking progress? It’s like adding five new things to your lifestyle in the hopes of bettering your X, and then always wondering which one made the difference. This gives “self-help” a bad name. There is no reason self-help should be pseudoscience.

It strikes me that this would also be a great way to teach kids statistics, too: perform long term experiments on themselves—batting average, weight, number of hours run in training for a race, and so on.