In the last post, a reader commented that the correlation would be more convincing if I graphed calories vs. average BMI rather than the prevalence of obesity. It was a valid point, so I went searching for average BMI values from NHANES surveys. I dug up a CDC document that contains data from surveys between 1960 and 2002 (1). Because these data only cover five survey periods, we only get five data points to analyze, as opposed to the eight used in the last post. The document contains BMI values for men and women separately, so I averaged the two to approximate average BMI in the general adult population. It's also worth noting that I use the approximate midpoint of the survey period as the year.
First, a graph of average BMI over time. It went up:
Now, let's see how well average BMI correlates with calorie intake:
The correlation between calorie intake and obesity prevalence was remarkable, but this correlation is simply incredible. An R-squared value of 0.98 indicates that daily calorie intake and average BMI are almost perfectly correlated.
We can further deduce that each 100-calorie increase in daily food intake is associated with an 0.62-point increase in average BMI among US adults.