Significant but not useful

When I was in graduate school, in the last millennium, I came up with the notion of "significant but not useful" to describe some data. I was doing research for a paper on radioprotective agents, and noted with interest that one agent that conferred protection was ginseng. The amount of protection conferred was significant – it was easily detected in the lab – but the effective change in the radiation dose was very small. The chance that even a very large dose of ginseng would make any real difference in the ability to withstand a radiation dose was quite small.

The difference was real, but not worth worrying about.

Here, we have the same sort of thing with cancer statistics.

Study confirms red meat link with bowel cancer People who eat more than 160 grams of red or processed meat a day are 35 percent more likely to develop bowel cancer than those who eat less than 20 grams a day, according to one of the biggest nutrition investigations ever carried out. [Story.]

Ok. 35 percent more likely than what? The study in question found 0.278% of the sample developed bowel cancer. If we take this figure as the baseline, we find that a 35% increase comes to a risk of 0.375% for those who eat a lot of read meat.

(160 grams is a shade under six ounces.)

If you essentially give up eating meat, you will decrease your overall chance of developing cancer by a tenth of a percentage point.

Whether it's worth it or not is up to you.