How many test subjects does it take to nail down a hypothesis?
When clinical researchers are planning a study, a least-favorite task is figuring out how many subjects are needed. The calculations get a bit esoteric, especially when “treatment” and “control” groups are of different sizes.
Dr. Michael Kohn of UCSF had a vision for an online suite of calculators for sample size (and related) calculations. He did the hard work of laying out the equations needed for the calculations. Mike Jarrett of QuesGen set up some early versions of several calculators, establishing a look and feel. I created a custom WordPress theme for the site, then got to work implementing calculators in PHP and Javascript, according to Michael’s equations. Like all my favorite projects, this one was kind of a brain-stretcher. (Ask me about Confidence Intervals some day if you want an earfull.)
The site is up and it’s doing a roaring trade! Check out the 17 featured calculators at sample-size.net. There’s good explanatory text and all the utilities are free, of course.
There are other plenty of online sample-size sites out there (some with beautiful graphics work), but we’re often told by researchers that they find our site the most complete and useful. Apparently the site is doing something right: with no promotion whatsoever we’re now up to 39,000 distinct visitors per month.
Leave a Reply