Unless you are clinically underweight or severely obese, junk food intake has no relationship to BMI

[eridani]

However, it probably impacts quality of life quite a bit.

http://www.npr.org/sections/thesalt/2015/11/12/455074815/are-junk-food-habits-driving-obesity-a-tale-of-two-studies

Let's start with the finding that seems most counterintuitive: For most of us, junk foods may not be what's driving weight gain. That's what behavioral economist David Just and his colleagues at the Cornell University Food and Brand Lab concluded in a paper in the journal Obesity Science & Practice.

The researchers looked at data collected in 2007-2008 from a nationally representative sample of roughly 5,000 U.S. adults as part of the National Health and Nutrition Examination Survey (NHANES), including information on weight, height and eating habits. Junk food was defined as fast food, soda and sweets.

Some of that data set had been used in a 2013 CDC study that found that heavier Americans were indeed getting more of their daily calories from fast food. But the Cornell researchers wondered what would happen if they excluded the people on the extreme ends of the weight spectrum — those who are clinically underweight and the very morbidly obese.

And they found that once those groups were eliminated, there was no association between body mass index and how much fast food, sugary sodas and sweets people consume.

Wondering if anyone has any clues to the answer to a statistical question. I have never been able to find any data on BMI except for averages, that is to say, the mean. This is a pretty useless number, as BMI statistical distribution is skewed. That absolutely must be the case because the typical ideal BMI of 25 cannot vary equally in higher or lower directions. You can only find someone with a BMI of 15 by digging with a good shovel. But there are people with BMIs of 35 or higher. If variation is not free in both directions, the skew must be toward heavier weights.

Which makes the mean the worst of all possible measures of the central tendency of a distribution. I did find one reference to the median (half above/half below) http://halls.md/body-mass-index-charts-men/ This is probably more connected to reality, but the only true measure of the central tendency of a skewed distribution is the mode, or most common value. I haven't been able to find a single reference about the BMI mode anywhere. Anyone have any ideas about where to look?