Design Template by Bikingtoronto

« It is On like Donkey Kong | Main | Virginia Closer to 3 feet to pass law »

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

Also Maryland got called out for being one of only a few states to have a mandatory bike lane use law.
I have to call out the callers out. Any state that has an "as far right as practicable on the roadway" law -- which is virtually all of them -- has a defacto mandatory bike lane law. Bike lanes are part of the roadway. I know, there are exceptions, but MD's bike lane law has exceptions.

The real problem is MD's mandatory shoulder use law. Bike lanes at least have some minimum standards. Shoulders aren't created with the idea that traffic will be using them and the law makes little distinction between usable and unusable ones.

I would guess farmers who live on the land they work would have no commute, so aren't counted. The bigger question is why is Alaska #1.

Running away from Grizzlies?

Just a thought, but it could be because generally speaking, cities in AK are not connected to other cities by road- and many of those cities are actually quite small. Add that gas in many places in AK is also quite high, and cars become much less useful than they are in the mainland.

It's also interesting to note that while biking and walking may beat car use in alaska, the state also has the highest number of airplane pilots per capita (about one in 78 residents has a pilot's license)

I'm not sure how they used the ACS to split out this data, but in the census factfinder, the categories that the ACS uses are:
- Walked
- Taxicab, motorcycle, bicycle, or other means

So are they just using that entire second category as "bicycle" and not separating it out?

Maybe those Alaskans are marking down "other" because they're riding snowmobiles to work...

As a Virginia resident, I would argue that Virginia deserves a low ranking on all things related to bicycling because the majority of state roads have no paved shoulders. Contrast this to New York where the shoulders are paved and wide.

With respect to the Alaska numbers, I wonder how people on oil platforms or fishing boats figure in? Do they self-report as walking (across the deck/platform) to work?

With respect to the graph, it does not prove coorelation or causation. Only statistical analysis would actually do that, this graph simply implies that there is a coorelation between the two.

1) No one said it proved correlation or causation.

2) I'll say it now, it does show a pretty strong correlation. There is an inverse relationship between the prevalence of active transportation and the rate of diabetes. You're correct that that does not prove causation.

3) Social scientists always get this wrong, because frankly they aren't really scientists - sorry. Correlation doesn't prove causation when there is no plausible link between two phenomena, or when there is some more plausible cause. But if there is a plausible link, then correlation is very strong evidence for causation. There are valid reasons to believe (research) that a community that bikes will have a lower rate of diabetes (hypothesis). The data above (experiment) seems to back up the hypothesis (conclusion). That's science.

3) Social scientists always get this wrong, because frankly they aren't really scientists - sorry. Correlation doesn't prove causation when there is no plausible link between two phenomena, or when there is some more plausible cause. But if there is a plausible link, then correlation is very strong evidence for causation. There are valid reasons to believe (research) that a community that bikes will have a lower rate of diabetes (hypothesis). The data above (experiment) seems to back up the hypothesis (conclusion). That's science.

You are still confusing correlation with causation. Yes there are "valid reasons to believe [...] that a community that bikes will have a lower rate of diabetes." All this requires is a basic correlation, which is clearly warranted by the data. If diabetes goes down when cycling goes up, then they are inversely correlated. That is what it *means* to be correlated. Both you and "guest" seem to be confused on this point.

The question is whether the cycling is the *cause* of the reduced rate of diabetes, which is an entirely different matter. In your discussion of the "plausible link" between cycling and lowered rates of diabetes, you fail to account for the possibility that two correlated phenomena may (and often do) correlate with a third, fourth, fifth, etc. phenomenon, each of which may provide another "plausible" explanation. This leaves you with multiple "plausible" explanations to choose from (and indeed more than one cause may be at work). Cycling, for instance, may correlate with diet, for example, or income and thus quality of health care. It takes careful analysis by trained statisticians to sort out these factors.

So no, correlation+ plausible link does not constitute "very strong" evidence. At best it constitutes a prima facie case.

Guez, perhaps I wasn't clear in my statement, but I'm not at all confused. I stated in part two that there is a correlation and noted that it was different than causation.

In part 3 I address the causation question. I agree that there are multiple possible explanations for the correlation (age for instance) and one of them may be more plausible (as I mentioned), but it is a reasonable conclusion, with the data given, that increased walking and cycling reduces diabetes (and frankly I think most doctors would agree). You're right that it is a prima facie case, which you say as though that is a bad thing. The point is that it stands until rebutted and guest made no rebuttal. Is it conclusive science? No. Is it strong evidence - I guess that depends on your opinion of strong.
But I stand behind my statement that when you have a plausible reason to believe that thing A causes thing B, and then it turns out that thing A and thing B are correlated in just the way you would expect if that were true; then that is strong evidence that your theory is valid. Pending further evidence it may not continue to stand. But it is, IMO, strong evidence.

"valid reasons to believe [...] that a community that bikes will have a lower rate of diabetes." All this requires is a basic correlation,

I see where you're confused. The "valid reasons" are not related to this graph. They are the wealth of data we already have that show an indisputable relationship between weight and diabetes and between weight and exercise. That's why I followed that part with the word "research" as that is an early step in the scientific process. So the correlation is not what we used to generate the hypothesis as that would be bad science. The correlation is the evidence that validates the hypothesis.

Washcycle,

I don't disagree with the assertion that cycling reduces diabetes, but I base that assumption on a vast body of peer-reviewed evidence that has been subjected to close statistical scrutiny. If all I had was this study, I would be skeptical.

For me, "strong evidence" would have to show that it was "more likely than not" that a causal link was present. Your formula doesn't meet this test for the reasons that I mention above: while it includes correlation and provides a plausible explanation, it does not adequately consider the possibly that there may be *better* "plausible" explanations.

guez: "Your formula doesn't meet this test for the reasons that I mention above: while it includes correlation and provides a plausible explanation, it does not adequately consider the possibly that there may be *better* "plausible" explanations. (emphasis mine)

But it does
Washcycle: Correlation doesn't prove causation when there is no plausible link between two phenomena, or when there is some more plausible cause. (emphasis mine again)

Washcycle,

You mention the possibility of a more plausible cause, but you don't adequately consider the implications. The bottom line is that until you have carefully *eliminated* other explanations, a plausible explanation is merely plausible. What makes an explanation strong is not being plausible, or being backed up by some sort of correlation, or both (as you suggest), but rather the fact that it is clearly better than the other available explanations. This chart does not meet that test.

I would agree that there is ample evidence elsewhere to suggest a causal link between exercise and reduced susceptibility to diabetes. That fact, however, doesn't make *this* evidence any less inconclusive.

Well, I'm not going to argue about what is strong evidence and what is just evidence. So I regret the use of the word "strong".

It just dawned on me that the "total number of trips" would include recreational trips, which would explain why the walking percentage is so much higher. IMO, a better number to focus on would be the number of work trips.

Obesity:

http://www.cdc.gov/obesity/images/map_county_obese_2007.jpg

Red States:

http://blogs.knoxnews.com/silence/2008-election-map.png

Here is a correlation between obesity and pick up driving rednecks. Strong scientific evidence suggests that obesity is linked with lower levels of education, and consequently Glenn Beck.

Glenn Beck is making people fat.

This is not inconclusive because I believe it to be so.

IMO, a better number to focus on would be the number of work trips.

Why? Do "work trips" cause less congestion or pollution?

Considering that peak hour/commuting trips are the main cause of congestion and the resultant pollution...

Any schmuck can go walk or bike around aimlessly for 30 minutes and it would count as part of the "total number of trips". If you truly want to make an impact, you want to capture those trips where people would otherwise use another mode. Namely work/commuting trips, but it'd also apply to shopping trips or others with a set origin/destination.

The comments to this entry are closed.

Banner design by creativecouchdesigns.com

City Paper's Best Local Bike Blog 2009

Categories

 Subscribe in a reader