I was looking through some data on how people's weight has changed over time in this CDC publication. I have been curious how researchers calculate age adjusted measures, so I wanted to delve into it a bit myself.
Here is an excerpt of the data for women from Table 6 in the CDC report.
It is apparent that weight has increased in every age range from the 1960-1962 time period (which I've called 1960) to 1999-2002 (which I've called 2000). It is also apparent that the age distribution has shifted, with a larger share of older women in 2000 than in 1960 (I'm assuming the sample sizes in these data are consistent with population statistics on age). So far, none of that is news. So, let's dig a little deeper.
The mean body weight of women in the US in 1960 can be calculated by multiplying the weight of women in each age range by the % of women in each age range and summing across all age ranges. When I do that, I find that the average weight of women in 1960 was 140.05 lbs.
Now, for a little thought thought experiment. What if we took the body weights by age observed in 1960 but instead assumed that we had the (older) age distribution that existed in 2000. To calculate this "strange" age-adjusted average, I multiply the weight of women in each age range in 1960 by the % in each age range in 2000 and sum across all age ranges. The result is 141.2 lbs. The difference is 1.15 lbs.
What does this mean? If the female population were, as a whole, older in 1960 - as it is today - then the average weight for women back then would have been 1.15 lbs higher. As a result, a small part of the weight gain from 1960 to 2000 (about 1.15 lbs worth) is due to changing age structure rather than weight gain per se.
The other thing I noticed is that it matters how you do the age adjustment. If I use the 1960 age distribution as the "base" then the age-adjusted weight gain from 1960-2000 is 24.1 lbs. However, if I use the 2000 age distribution as the "base" then the age-adjusted weight gain from 1960-2000 is "only" 23.1 lbs. Thus, we can push our weight gain figure up or down a pound simply by choosing which base we wish to use.
It is interesting to note that weight gains have been highest among the youngest females and lowest among the oldest. The gain in weight from 1960 to 2000 for 20-29 year olds was 28.8 lbs but the gain for 60-74 year olds was only 17.4 lbs. Also, the variation across years has changed a lot. In 1960, the difference in 20-29 year olds and 60-74 year olds was 19.6 lbs. Today, that same difference is only 8.2 lbs. So, we are fatter but more equal?
Regardless of what has happened to weight, I can't help but think that it is a very good thing that we have more older women in 2000 than 1960. Sure, they're a little fatter, but probably thankful to still be kicking.
P.S. It is important to note that the gains from 1960 to 2000 have, in the recent decade, leveled off and there does not appear to be much change in obesity prevalence among women in the past 10 years (see this paper in the Journal of the American Medical Association or my previous blog post).