This is the greeting from Steve Halls, MD, at his web weight-calculator. After a fair amount of searching on the internet, I found this site on body-mass the easiest to use and informative. However, I cannot speak on its accuracy. I will only admit that it provided far less scary news (and realistic, I feel) about my own weight than other websites giving advice on this vital subject.
According to the “updated hall.md v2” standards, I am “marginally overweight” at the 53rd percentile of other American males at my age and height. As we like to say in Minnesota, this could be worse, so it’s not so bad.
Discussing what should be the “ideal” weight would take up a great deal of time and energy: Never mind that. What I want to do is focus on monitoring weight. For example, I just completed the pictured outlier-detecting run-chart* on my 20 weighings** thus far this year. Notice that none of the results fall outside of the 95 percent confidence limits.
Even so, after I penciled in my number for the highlighted point, my wife hassled me a bit about going overweight when she saw . I predicted that she would see a regression to the mean, which didn’t impress her one bit. Nevertheless, the value of being patient by charting data over a period of time can be seen in this instance – it vindicates me not reacting to one result.
Coincidentally, our contract trainer Doug Hubbell came to Minneapolis for our new Advanced Formulations workshop. He is the author of a handbook for managers seeking quality improvement (Managing for Profits – to be published soon). Doug is a plain-talking straight-shooter who rifles in on what’s needed to stop chronic manufacturing waste. Charting is a powerful part of his arsenal of quality tools. His reaction to me mentioning my monitoring of weight was “I hope you do not expect this chart to help you lose pounds.” Naturally I wouldn’t admit to that, but, honestly, it did cross my (hopeful!) mind. However, I am mainly just trying to track a very gradual increase of about 1 pound per year since my high-school graduation, when I was in the best shape of my life.
The battle against the bulge continues…
*Using Design-Expert® software’s diagnostics tools. I focused on a chart that deletes each point before calculating its deviation in terms of standard deviation, which makes it more sensitive to statistical outliers. For details, see this Wikipedia entry on Studentized residual (it explains internal and external methods).
**Done with a new bathroom scale that I really like – this Precision Digital model by EatSmart.
#1 by Wayne on February 28, 2011 - 3:54 pm
What is more likely? Is weight not a function of time, or is it cyclic through time? There’s a quadratic peak early in time and another late?
A designed experiment should be done one run at a time. Measure the response at the specified time; then reset the process and run for its appointed time.
If this is continuous, over time, same starting conditions, time series, data mining, then time series specialized tools will serve better.