A 70% Chance-- Or 4 Times As Likely---

[Always Free]

to be hurt in a car accident without a seat belt than if you wore a seat belt, or to get lung cancer if you smoke than if you don't smoke???

This is not a real statement but just an example--but----

What does it actually mean---4 times as likely? or 70% chance of?

You hear people refer to this all the time. "If you don't eat the right food and gain weight and don't excersise you are 6 times more likely to become diabetic."

How do they work out the '6 times' bit? Why is it not 20 times or 40 times more likely that you will become diabetic?

Could it be that 6 out of 10 people actually get diabeties if they are overweight there for you are 6 times more likely etc etc?

Could it be that of all people who are in a car accident who don't have a seat belt on, 70% of them will be hurt?

Is this how these figures are arrived at? I can't work it out.

 

[arcticpenguin]

You seem to be just making up numbers. Please pick one example, and provide a link for it.

Thank you,

 

[Paul C. Anagnostopoulos]

Yes, a specific example would be nice.

I presume it's a simple thing. Among people with blonde hair, 7 in 100,000 are afflicted with snorkiosis. Among people with dark hair, 42 in 100,000 are afflicted. Dark-haired people are 6 times as likely to have snorkiosis as light-haired people.

Of course, if you read this in a newspaper, it's certain to be incorrect, because it involves numbers.

~~ Paul

 

[Always Free]

I did make up the numbers as I stated.

Do you mean to say that you have never heard anyone say something like---If you smoke you are 'so many' times likely to get lung cancer than someone who doesn't smoke? Or how about if you use too much salt on your food you have a 'certain percent' chance of getting high blood pressure than if you don't overdo your salt intake.

Sheesh, I knew people would have trouble understanding exactly what it is I am asking.

I'm not trying to find out whether the figures given for, lets say contracting lung cancer if you smoke compared to if you don't smoke, are scientifically correct. I'm trying to figure out what does---'so many times likely' or 'a certain percent chance of' actually mean.
Should I worry about doing a certain thing if someone says it is 5 times more likely to harm me or should I only be worried if the figure is--let's say--43 times more likely to harm me?

 

[arcticpenguin]

The answers depemd on the specifics.

For example, if something is 5 times more likely to harm you, what are the actual numbers? Is your chance 1/10 000 000 000 if you do thing A and 5/10 000 000 000 if you do activity B?

 

[tracer]

Among people with blonde hair, 7 in 100,000 are afflicted with snorkiosis. Among people with dark hair, 42 in 100,000 are afflicted. Dark-haired people are 6 times as likely to have snorkiosis as light-haired people.

Please, won't you give generously to the American Snorkiosis Foundation?

 

[Dancing David]

There are two ways that the numbers are generated.

the first is to plant math seeds and water them daily and then with careful supervision they becaome math plant, if fertilized the they bear fruits, alos called statistics.

The second is generaly three fold.
There are actuarial statistics where for the general population a tend is noticed. Paul's example of snorkorsis.
Then there are longitudinal studies, a population is devided into two population, those that use xenpfrebe and those that don't, if it turns out that there is a seventy percent mortality compared to those who don't xenefrebe then there is the seventy percent more likely.
More logitudinal, like the nurses studie show that x cause 20 while y causes 40, this can be twice as likely or fifty percent more likely.

I have found that yahoo or other news sites will give links to more analysis.

 

[Aoidoi]

Not sure if this is what you're looking for, but here's how I think of what the statistics mean:

Pick a random person Joe. If Joe is a non-smoker the chances that he will get lung cancer at some point is x (say, .1%. If he smokes it's 6x (say, .6%).

As to avoiding something that makes an outcome the x times more likely, well, that depends on the effect. If it increases your rate of heart disease by 10x, I'd avoid it because heart disease is a leading cause of death. If it increases your odds of getting some peculiar rare form of big toe cancer of which there are only a hundred known cases throughout history I wouldn't be worried.

To put that more mathematically: If your chance of dying from heart disease is normally 10%, but bathing in margarine triples the risk then you're chance of dying of heart disease are 30%. If your normal chances of dying of big toe cancer are .000001% and playing soccer increases your odds 3x your chances are then .000003% (which is still negligible). So it depends on both the increase in chance as well as the likelyhood of it occuring.

Increasing your chances by 70% would, I think, mean that you are 1.7 times as likely to have it happen (and then calculate as above). Depends on how they phrase it.

As to how to collect the statistics, well, David gave a bunch of info on that. Doing it well is a pain in the big toe, but what you usually see in headlines (Broccoli decreases cancer risk by 5%) is generally not done well and is useless as a criteria for making choices.

Feel free to correct any incorrect info provided, I tend to get my brain headed in the wrong direction with stats on a fairly regular basis.

 

[tracer]

To put that more mathematically: If your chance of dying from heart disease is normally 10%, but bathing in margarine triples the risk then you're chance of dying of heart disease are 30%.

Even then, you have to be careful not to confuse correlation with causation. Does bathing in margarine itself really triple your chances of getting heart disease, or is it merely that among your representative sample groups, the percentage of the people who happened to bathe in margarine and also happened to get heart disease was 3x higher than the percentage of people who didn't happen to bathe in margarine and also happened to get heart disease? The practice of margarine bathing might not be causatory -- it might only be an indicator of some other factor among people who have a propensity for heart disease. (Maybe the heart disease gene also gives you a desire to take margarine baths, for example.)

The only way to distinguish between correlation and causation would be to randomly divide your subjects into two groups, then force group A to bathe in margarine and force group B not to bathe in margarine. Then compare their rates of heart disease.

 

[Aoidoi]

The only way to distinguish between correlation and causation would be to randomly divide your subjects into two groups, then force group A to bathe in margarine and force group B not to bathe in margarine. Then compare their rates of heart disease.

A very good point. And due to the difficulties of conducting a lot of research like that on humans there are often questions that linger regarding causality. (The ethics of human experimentation get messy real quick. How'd you like to selected for the bathe in margarine group? )

What that means for an individual, I suppose, is that changing your habits might not help, but the stats give you a better idea on what will kill you (i.e. if you're already bathing in margarine, you know you're 3 times likelier to die, but if you stop it might not drop your rate down to normal.)

 

[Dancing David]

The other issue in studies is that most people won't get help until the last moment. Most of the placebo effect is attributable to regression to the mean.