Being a spoilsport

For the last few hours, Yahoo! decided that I'd be interested in reading this piece of news. Every time I go there, I get this front page:

 

I don't really know how this sort of studies gets published in journals, nor am I interested in spending an hour figuring out how they failed to prove it. The snippet summarizing the research is here.

The ability to look at some data and develop some sense of whether it makes sense is an important practical skill, especially these days when there is so much data being brandied around. There are quite a few things we already know without reading the research. This is not like a cholerstorol screening test for example, in which if the test shows high, there is medicine and diet changes that could bring down the level, and thus potentially prolong one's longevity. What is proposed here as a test does not lend itself to any kind of remedy. Also, I'm not sure if they proved predictive ability, or just proved a correlation. Sounds like the latter. Finally, even if true, my guess is that the death rate of the population they studied, aged 51-80, has relatively few deaths in a six-year follow-up period, making even small differences look huge on a ratio scale.

***

For those who don't have time to click, the test that has a 6.5 times (!!!) lift in predictive accuracy involves sitting and standing. Sorry to spoil your dreams of eternal life.

Comments

[Trent McBride]

It gets better:

"More than half the participants ages 76 to 80 failed the tests, scoring 0 to 3. Not surprising around 70 percent of those under 60 earned a near perfect or perfect score of 8, 9, or 10."

I wonder if could have gotten the same predictive power just by asking age and foregone all the sitting nonsense.

[Kaiser]

I was thinking the same thing this morning. Is it better than looking at an actuarial table?

[Floormaster Squeeze]

I like the image of data (and poor analysis) being "brandied around".