Item — HN Reader

Although this is an interesting result, beware, n=25 for the whole study, which means there were only eight people per group.

We did a data analysis on several thousand Apple Watch users, looking at whether step count or exercise intensity really drives one key outcome (a lower resting heart rate):

http://blog.cardiogr.am/2016/02/12/do-you-really-need-10000-...

Like this study, our conclusion was that intensity matters most: you need to get your heart rate above 150 to get the best benefit. In our data set, 45 minutes of exercise per week was the "knee" of the curve.

I'd actually love to run a scaled-up version of the NYT study. I think with wearables generating so much data, we really can start to answer questions like: "What is the best workout?" Or perhaps more importantly, "What is the best workout for you?"

1 comments

> Although this is an interesting result, beware, n=25 for the whole study, which means there were only eight people per group.

I'm so tired of this comment being made about every single study that finds itself on HN. It's a complete misunderstanding about how statistics work. The whole point is to use a small sample to say that something about the larger population.

Pardon my ignorance, as my memory is a little foggy when it comes to statistics and scientific studies, but doesn't the sample size need to be large enough to represent the population? I forget exactly how that works.

Yes, but virtually every published peer-reviewed study will have a large enough sample size and will have done the math to make sure that it is large enough.

If it's the first thing that comes to the mind of random HN commenters who haven't even read the actual article, it's probably also going to occur to the experts writing and reviewing the article.

Peer reviewed articles aren't perfect, but it's pretty arrogant to think that the authors and reviewers are so bad at research that you can find problems with them without even reading them.