tag:blogger.com,1999:blog-8566036031670722613.post1507768278941679373..comments2018-11-06T00:48:53.265-08:00Comments on Graph of the Week: Body Weight in the United States - Part 1, "The Problem"Patrick Rhodeshttp://www.blogger.com/profile/14874894005290887213noreply@blogger.comBlogger5125tag:blogger.com,1999:blog-8566036031670722613.post-46090577339200920042012-06-14T11:22:55.322-07:002012-06-14T11:22:55.322-07:00Good info.
Regarding the first graph, I didn'...Good info.<br /><br />Regarding the first graph, I didn't 'scale' anything. I cropped the area to show the relevant data - completely different than scaling. I try very hard *not* to produce misleading graphs - there are enough of those in mainstream media. :-)<br /><br />Thanks for your comments.dezert1https://www.blogger.com/profile/01201701558271990438noreply@blogger.comtag:blogger.com,1999:blog-8566036031670722613.post-82799706741548794592012-06-13T10:48:06.276-07:002012-06-13T10:48:06.276-07:00There _is_ an allometric effect, and your linear f...There _is_ an allometric effect, and your linear fit would be dismissed by anyone who deals with obesity in prevalence studies.<br /> <br />By Patrick's reasoning (linear relationship), either:<br />18" 7 lbs<br />double: 36" = 14 lbs<br />quadruple: 72" = 28 lbs<br />or, if you keep your averages from adults, your 2.75 lb/" would have that 18" newborn weigh 49.5 lbs (ouch!).<br /><br />If the shape was similar (a cube for simplicity, or a baby scaled up to a pudgy adult), the volume (and thus mass) would scale with the cube (^3) of a length: doubling the height would mean 8x the weight. <br />18" to 36" = 2x so weight would be 7 * 8 = 56<br />18" to 72" = 4x so 7 * 64 = 448, even greater than 343<br />But teenage let alone adult humans are not pudgy shaped like infants: as humans grow, they become more elongated, so the empirical exponent is less than 3 (the empirical exponent differs among species).<br /><br />For adults humans, the standard correction to make a scalar obesity index (Quetelet, often BMI) is weight / height^2 (using kg and cm, or multiplied by 703 if you use lbs & inches); the exponent of 2 is pretty close to the empirical exponent. The major advantage of using Quetelets for your comparison is that they do a better job of capturing the variation among people at a given time: the BMI calculated form the average weight and average height will not be the same as the average of BMIs calculated on each individual. <br /><br />If anyone finds this post while searching for obesity, I encourage you to look at how the field deals with scaling weight for height.<br /><br />On a graph style note, given that your Y axis are percentages that add to 100, I think that your first graph would be less misleading if the Y axis ran from 0 to 50. The way you scaled it, the visual impression is that in 1995, the prevalence of obese is roughly 1/10th of overweight.tomphttps://www.blogger.com/profile/01237658207649473532noreply@blogger.comtag:blogger.com,1999:blog-8566036031670722613.post-31425893354512441462012-06-09T00:43:27.543-07:002012-06-09T00:43:27.543-07:00yeah - you're right... thanks for the reply!yeah - you're right... thanks for the reply!Anonymoushttps://www.blogger.com/profile/03861371577048804644noreply@blogger.comtag:blogger.com,1999:blog-8566036031670722613.post-56849265607029606622012-06-08T13:20:04.892-07:002012-06-08T13:20:04.892-07:00I'm not sure the allometric effect works here ...I'm not sure the allometric effect works here since we're talking about the same species (humans). Using your argument, if a newborn baby is 18 inches tall and weighs 7 lbs (average figures), then by the time they are 36 inches tall (3 feet), they should weight 49 lbs - okay there.<br /><br />BUT - assuming they reach 72 inches (6 feet), by your reasoning they should weight 343 lbs!<br /><br />So, while there might be a slight allometric effect in play, it certainly isn't enough to worry about here and will not make up the 22 lb difference mentioned in the article.dezert1https://www.blogger.com/profile/01201701558271990438noreply@blogger.comtag:blogger.com,1999:blog-8566036031670722613.post-83660768843574181022012-06-07T00:14:14.143-07:002012-06-07T00:14:14.143-07:00I think your calculation of weight per height is n...I think your calculation of weight per height is not correct. There is a allometric effect you should take in account: Doubled height means squared volumina - therefore weight-gain must be more than linear.Anonymoushttps://www.blogger.com/profile/03861371577048804644noreply@blogger.com