3 ms·

Looks like a pretty classic correlation != causation issue. Yeah, may be true what they found. But may also be not very relevant. It absoltely doesn't say that

by hannob 7y ago

Looks like a pretty classic correlation != causation issue. Yeah, may be true what they found. But may also be not very relevant. It absoltely doesn't say that getting an e-bike will make you exercise more. Here's a theory what might be going on: E-Bikes are relatively expensive. So if you buy an E-Bike you're probably relatively determined to actually use it. There are on the other hand plenty of people that own a cheap bike and use it very rarely. There's no real reason not to do a proper study on the question: Take a reasonably large number of volunteers, randomly assign them to "we give you a fancy bike" or "we give you a fancy e-bike" group and then measure their exercise and health levels. There are situations where doing such a randomized trial is unethical or pracitcally impossible. Here it really is not. If someone's doing a study with observational data where you could do a randomized trial you should be skeptical.

dmurray 7y ago
Your trial sounds expensive, maybe 50 times as expensive as the version where you just talk to people, for the same statistical power. So that's at least one reason not to do it. And you still end up measuring something different from what you want. Just as in your theory about people committing to use their e-bike more when they've spent a lot of money on it, people behave differently when you give them things for free.
- hannob 7y ago
  > Your trial sounds expensive, maybe 50 times as expensive as the version where you just talk to people, for the same statistical power. It is, especially if you do it with proper statistical power. Though I'm deeply convinced that it would do science a lot of good to do fewer more expensive, but meaningful trials instead of more meaningless ones. Your concern is somewhat valid, as you can't "blind" the test, but by using similarly priced bikes you could at least cancel out a price bias. Still there might be a "new tech" bias towards e-bikes. But then we're at "this is the best we can do" and there are limitations of how good we can study things. The study quoted above is not even near trying to do that.
usrusr 7y ago
"E-Bikes are relatively expensive. So if you buy an E-Bike you're probably relatively determined to actually use it." Seems like you have the blueprint for the "perfect" counter-study: compare e-bikers only to people who spent about as much on a conventional bicycle and you select for moderate fitness nerds who will routinely output more power than e-bike and e-biker together and spend many hours per week training. The results will go through the roof in favor of conventional bikes! Utterly skewed, but nominally apples to apples.
GhostVII 7y ago
I think the simpler explanation is that the most common way people get access to an ebike is through things like Jump bike, and people who use those bike sharing services are often using it to commute places. They will probably end up spending far more time biking than someone who drives to work, but has a bike they use on the weekend recreationally.

by hannob 7y ago

dmurray 7y ago

Your trial sounds expensive, maybe 50 times as expensive as the version where you just talk to people, for the same statistical power. So that's at least one reason not to do it. And you still end up measuring something different from what you want. Just as in your theory about people committing to use their e-bike more when they've spent a lot of money on it, people behave differently when you give them things for free.

hannob 7y ago
> Your trial sounds expensive, maybe 50 times as expensive as the version where you just talk to people, for the same statistical power. It is, especially if you do it with proper statistical power. Though I'm deeply convinced that it would do science a lot of good to do fewer more expensive, but meaningful trials instead of more meaningless ones. Your concern is somewhat valid, as you can't "blind" the test, but by using similarly priced bikes you could at least cancel out a price bias. Still there might be a "new tech" bias towards e-bikes. But then we're at "this is the best we can do" and there are limitations of how good we can study things. The study quoted above is not even near trying to do that.

usrusr 7y ago

"E-Bikes are relatively expensive. So if you buy an E-Bike you're probably relatively determined to actually use it." Seems like you have the blueprint for the "perfect" counter-study: compare e-bikers only to people who spent about as much on a conventional bicycle and you select for moderate fitness nerds who will routinely output more power than e-bike and e-biker together and spend many hours per week training. The results will go through the roof in favor of conventional bikes! Utterly skewed, but nominally apples to apples.

GhostVII 7y ago

I think the simpler explanation is that the most common way people get access to an ebike is through things like Jump bike, and people who use those bike sharing services are often using it to commute places. They will probably end up spending far more time biking than someone who drives to work, but has a bike they use on the weekend recreationally.