So many times when we consider numbers, the numbers are proxy variables for something else. Yet, numbers are often misleading. As a famous phrase goes, “There are lies, damned lies, and statistics.” And over the last few days, I’ve found a few things that make the variable of steps to be a proxy.
With the popularity of fitness watches (e.g. Fitbit, Garmin vivofit, etc.), the recommendation of 10,000 steps in a day is often a goal that people try to set. However, the steps themselves might be a proxy variable.
How is it a proxy variable? There are several reasons:
- Upper body exercises. If you’re doing upper-body-strength workouts, such as weight-lifting or other resistance exercises, most of the exercise will not involve taking many steps, but it is still healthy to work out your upper body. Or, yoga might be another example of a good exercise that doesn’t generate steps.
- Bike riding. I remember when I was on my second Trail Trek back in 2004, and I wore a pedometer while riding. After 24 miles of riding, it only registered 1000 steps. I had certainly pumped my legs more than 1000 times, because I need more than 20 revolutions in order to ride a mile on a non-downhill!
Even if you are using a fitness watch, the steps remain a proxy variable, because the motion of pedaling is slightly different than the motion of walking (especially if the watch reacts more to arm AND leg movement).
- Running. Most people have a longer stride length when they run versus when they walk. I found this out informally last week when I took a “sanity walk” around my academic building. It took about 1000 steps and about 8 minutes.
Later that day, I did the same course, but jogging. It took about 700 steps and 4 minutes.
Granted, this meant that the derivative had a larger magnitude, despite the fact that the function value was lower. (Sorry for throwing calculus into that one :p).
The idea of getting 10,000 steps in a day is certainly a good guideline, but as much as any other guideline, you have to critically evaluate it and realize when your variables might be proxy variables, or when the guideline doesn’t really apply to your situation.
What other numbers can you think of which are proxies for something qualitative?
Columbus: 2 days.
Twenty-Nine: 5 days.
Kenosha: 32 days.