Followup to Barefoot Running Efficiency Study

I got asked a number of questions about my previous look at the Running Shod More Efficient study, so I thought I’d answer them here.

I also took another look at the study itself, and managed to ask myself a question that I sure couldn’t answer.

Why did they control for factors that aren’t controlled in the real world?

As noted in the study, there have been seven previous studies, and five of those showed no difference between energy use while shod or barefoot. What these folks tried to do is tease out if there were any differences. They really wanted to ask the question about whether running barefoot made a difference, not whether, for instance, running with a midfoot-strike versus running with a heel-strike made a difference.

In any good study, you try to keep as much the same as possible.

Why run on a treadmill instead of normal running conditions?

It’s easier to control the speed.

I do agree that there might be conditions “overground” that could make a difference, but this study just didn’t test that. On the other hand, there have been other studies that compare running on treadmills with running overground. It turns out that those studies find that they are pretty much equivalent (at least for the conditions of those tests) if you tilt the treadmill to a 1% incline (to compensate for lack of wind resistance).

The previous overground study (the one by Hanson) probably screwed that up by relying on a speed measuring device that did not adjust to the difference in stride length between running barefoot and running shod.

I still would have liked to have seen a decent test overground. I think they could have collected the data they needed by filming it and analyzing those pictures.

What about that stupid yoga sock?

My suspicion is that it really didn’t make a difference. It may have—we just don’t know.

I do think it is a flaw in their design, though.

Why use all the different weights? Why not just test barefoot versus shod?

Here’s the data of simply barefoot versus shod with that very lightweight shoe. (From what I can tell, the shoe they used weighs about half the weight of a typical running shoe.)

The "Pure" "Barefoot" versus Shod Graph

There really is no significant difference there. And even their conclusion (where they compared shod with the light-weight shoes with weighted “bare” feet) had overlapping error bars, which tells me that the results really aren’t significant. Yet, they claim a (p<0.05) significance to that.

I still think that the difference seen is completely related to the difference in stride lengths between their shod and "barefoot" runners. Remember, the barefoot runners used a 3% shorter stride.

Using the different weights actually allowed a check with previous studies, which had found about a 1% increase in metabolic effort with each 100 grams of weight. These folks found the same thing, which leads credence to their results.

Why would a shorter stride length lead to higher effort? Look at cyclists. They have an incredibly high turnover and that’s considered efficient.

To go the same speed with a shorter stride requires a higher cadence. With cycling you have gears to compensate for and take advantage of that higher cadence, keeping your body’s efficiency in that “sweet spot”. I don’t think the same applies with running (but I could be wrong).

As an extreme example, if you run in place, you’re still expending about the same energy but with zero speed. Most of the energy is expended during landing and take-off phase, so if you do that more often (in order to maintain a higher speed with a shorter stride), you will probably burn more energy.

And here’s my question: what were the stride lengths, and do they make any sense?

According to the data in the study, the stride length when “barefoot” was 2.18 meters (7ft, 1.8in), and 2.24 meters (7ft 4.2in) when shod. (Don’t forget that a “stride” is the distance from one foot hitting to when the same foot hits again, so the respective step lengths were 3ft 6.9in and 3ft 8.1in.)

This could still implicate those stupid yoga socks. We don’t know if the “barefoot” runners had a shorter stride because of running without a sole, or because there was some effect due to slippage of the yoga socks on the windmill.

But what about my unanswered question?

The paper reports the stride length as 2.18 meters with a standard deviation of 0.59 meters (“barefoot”) and 2.24 meters with a standard deviation of 0.55 meters. Those are huge standard deviations, nearly 22 inches for the stride length (11 inches step length).

Now, the way a standard deviation works 68% of the sample should fall within one standard deviation (and 95% within two standard deviations). Going with just one standard deviation, that means that they had people with step lengths of 2ft 7in and 4ft 6in. And if you go to two standard deviations, . . ., wow. They must have been testing giants and dwarfs.

How the heck did they get these results? I have no idea.

The more I think about it, the more I think they combined their data incorrectly. Instead of just combining the results from all their participants, they should have measured and analyzed the results for each individual separately, and then figured out how to combine the data. This way, we have no idea how each individual performed—all we see is the mash-up.

This also makes their other results suspect, and now that I look at it, I am even more confused about what they are claiming. The error bars on a plot are supposed to show one standard deviation. If you look at the “barefoot” point above (no weight), that looks like a value of 40.3 with an error bar of about 1.

What’s the value in their paper?

40.28 with a standard deviation of 3.05.

Here’s the screen capture of their table:

The Data Values Table (click for readable version)

I don’t see how these results can be significant at all!

End of questions; back to commentary.

But let’s suppose that the results really are significant. What does it mean to us?

As I mentioned before, it probably only makes a difference if you are an elite athlete really trying to shave off that last millisecond. Otherwise (and as others have pointed out) all the other variables will totally overwhelm this particular point.

There is something else going on here: many of us are indignant that bare feet might be worse, and we are working to find any reason to doubt the study. I feel the lure myself.

We want to think that bare feet are the greatest thing since sliced bread! We’ve found the cure! It’s amazing! It slices, it dices . . .

Seriously, going barefoot doesn’t have to be best in everything. It just has to be best for us.

One more point.

When driving above 40 miles per hour, closing your car windows and using the air conditioner gives better gas mileage. But if it is the first gorgeous spring day after a hard winter, and the temperature is 75°, you’re going to drive around with your windows open. The heck with efficiency.

And when you are barefoot, it is always a beautiful spring day . . .