How do I lose 5 pounds in 7 days?

[DGuller]

The 2nd one lists that those with normal BMIs were more at risk to die than overweight BMIs when both had the same pre-existing condition.

The key is the risk wasn't the same.  The only bad result in statistics is the independence of predictive variables and outcomes.  If having speeding tickets was correlated with having less accidents, knowing whether you have speeding tickets would still be just as important.

The 2nd shows that it wasn't correlated with several causes of mortality nor mortality itself.

I'm not sure where it says that.

I don't think the point is to show that it has no use but that its flawed enough that some other model should be (or variables) taken into account.

If that was your point, then it doesn't address the rebuke of merithyn's argument.  She claims that BMI is completely useless, and your quote doesn't support that assertion.

[garbon]

The key is the risk wasn't the same.  The only bad result in statistics is the independence of predictive variables and outcomes.  If having speeding tickets was correlated with having less accidents, knowing whether you have speeding tickets would still be just as important.

So are these hypothetical insurance companies charging more to normal BMI individuals because we know if they develop cardiovascular disease they have a higher risk of dying?

I'm not sure where it says that.

In the bit where it says it is not a good measure (and I'd guess predictor) of several causes of mortality / mortality.

If that was your point, then it doesn't address the rebuke of merithyn's argument.  She claims that BMI is completely useless, and your quote doesn't support that assertion.

Actually didn't she say something about it being useful (or at least used) for populations in general? Her complaint was about applying the heuristic to specific individuals.

[derspiess]

My point of contention is that if it isn't actually correlated with a lot of causes of mortality, it doesn't seem particularly useful.

BMI is not correlated with health problems?

How about what you re-read what I wrote?

Okay, then how is mortality relevant unless we're talking about life insurance?

[DGuller]

So are these hypothetical insurance companies charging more to normal BMI individuals because we know if they develop cardiovascular disease they have a higher risk of dying?

I don't know, and that's not particularly relevant.  Even if insurance companies are mispricing people on BMI, it's not really the fault of BMI.

In the bit where it says it is not a good measure (and I'd guess predictor) of several causes of mortality / mortality.

I believe I already covered the "imperfect equals useless" fallacy.

Actually didn't she say something about it being useful (or at least used) for populations in general? Her complaint was about applying the heuristic to specific individuals.

Yes, but that's a distinction without a difference.  Something that is predictive for populations is predictive for individuals as well.

[garbon]

I don't know, and that's not particularly relevant.  Even if insurance companies are mispricing people on BMI, it's not really the fault of BMI.

How is it not relevant? The discussion is about whether insurance companies should be pricing based on BMI.

I believe I already covered the "imperfect equals useless" fallacy.

Not relevant for what I'm saying though as I haven't maintained that BMI is completely useless but rather that it may be flawed enough to not be a good stand-in for health events that individuals will suffer.

Yes, but that's a distinction without a difference.  Something that is predictive for populations is predictive for individuals as well.

I'm not sure. After all it is specific individuals being insured. What if a given carrier's population isn't representive of general population as whole and instead is skewed towards individuals for whom BMI is a poor measure?

[garbon]

Okay, then how is mortality relevant unless we're talking about life insurance?

Well perhaps it is a big assumption on my part, but I would think mortality would be closely linked with potential health costs that an individual might incur as many cases of mortality will stem from health conditions.

[Malthus]

It's a really poor predictor when applied to the individual. It works for the insurance company because they are dealing with thousands of people. It sucks for the atypical individual because they are not.

So is every single other rating variable.

Right, which is why if you want to be more fair to the individual, you need to collect lots of data. Which means having the insurer take a greater interest in your personal data, meaning more intrusion (and some higher costs).

Is the fairness worth it? Some people built like wrestlers will get a better worse deal it is true.

FYP since wrestlers are not likely to be in the "healthy" part of the BMI.

Actually, even using the old standard from the 1980s - don't remember what it was called, but the scale included height, weight, and body type - would be a HUGE step up from the BMI.

DGuller, you seem to be advocating to allow an imperfect measurement simply because, well, nothing's perfect. Just because nothing is perfect doesn't mean that there can't be a better means found. I'm advocating for that better means, not to disregard all data.

What I mean is that if a system were more fair, people built like wrestlers would get a better deal. They are (presumably) in the "bad" BMI range but are, in fact, healthy. 

The thing is that generally, the more fair the system, the more intrusive the company has to be. You mentioned calipers upthread. That sort of thing.

[The Larch]

It's a really poor predictor when applied to the individual. It works for the insurance company because they are dealing with thousands of people. It sucks for the atypical individual because they are not.

So is every single other rating variable.

Right, which is why if you want to be more fair to the individual, you need to collect lots of data. Which means having the insurer take a greater interest in your personal data, meaning more intrusion (and some higher costs).

Is the fairness worth it? Some people built like wrestlers will get a better worse deal it is true.

FYP since wrestlers are not likely to be in the "healthy" part of the BMI.

Actually, even using the old standard from the 1980s - don't remember what it was called, but the scale included height, weight, and body type - would be a HUGE step up from the BMI.

DGuller, you seem to be advocating to allow an imperfect measurement simply because, well, nothing's perfect. Just because nothing is perfect doesn't mean that there can't be a better means found. I'm advocating for that better means, not to disregard all data.

What I mean is that if a system were more fair, people built like wrestlers would get a better deal. They are (presumably) in the "bad" BMI range but are, in fact, healthy. 

The thing is that generally, the more fair the system, the more intrusive the company has to be. You mentioned calipers upthread. That sort of thing.

What's the point of a non-intrusive physical check for health insurance? There are good reasons for the information to be collected.

[garbon]

The thing is that generally, the more fair the system, the more intrusive the company has to be. You mentioned calipers upthread. That sort of thing.

I think that's the crux of it and what I've no idea about. How much of our medical history/info do our insurance companies already have?

[Malthus]

What's the point of a non-intrusive physical check for health insurance? There are good reasons for the information to be collected.

Well, that's just the issue. Allegedly, they are presently basing the decision purely on BMI, which is relatively non-intrusive.

A "better" as in fairer system would require more thoroughness. But there costs involved in being more throrough. The issue is whether is is better to use a cruder, more infair system that is cheap, or a mores sophisticated system that is also more expensive (both in money and in annoyance)

[MadImmortalMan]

They're already spending the time and money to do a physical.

[Malthus]

They're already spending the time and money to do a physical.

According to the OP, the rates will be set by BMI alone, which indicates that the "physical" is pretty superficial. If not, it is admittedly a weird system, to do a thorough physical and then base the rates on BMI only.

[merithyn]

They're already spending the time and money to do a physical.

According to the OP, the rates will be set by BMI alone, which indicates that the "physical" is pretty superficial. If not, it is admittedly a weird system, to do a thorough physical and then base the rates on BMI only.

And yet, that is exactly how it works. I have to have a complete physical with blood work and TB test, but the increase is based entirely on the BMI.

[Malthus]

They're already spending the time and money to do a physical.

According to the OP, the rates will be set by BMI alone, which indicates that the "physical" is pretty superficial. If not, it is admittedly a weird system, to do a thorough physical and then base the rates on BMI only.

And yet, that is exactly how it works. I have to have a complete physical with blood work and TB test, but the increase is based entirely on the BMI.

In that case, the system is indeed silly, as they already have the info to make a better determination. What's the rest of the physical for - to determine whether to screen you out from getting insurance at all?

[MadImmortalMan]

Yeah probably. HIV positive?

Oh, uh...suddenly we found a better applicant...