[margalabargala]

I would expect it to be the other way around.

If nearly everyone smoked, then even nonsmokers were constantly getting a fair amount of secondhand smoke.

This would raise the background rate of cancer, making it appear that smoking raises your risk by less than it actually does.

[rjsw]

Non smokers did get lung cancer [0].

[0] https://en.wikipedia.org/wiki/Roy_Castle#Illness_and_death

[margalabargala]

Yes, exactly my point.

If the "normal" rate of lung cancer is X, the observed rate in nonsmokers who get secondhand smoke is X+Y, and the observed rate in smokers is X+Y+Z, if you compare nonsmokers and smokers it looks like smoking increases your rate by Z when it's actually Y+Z.

[paulpauper]

this agrees with my point because non-smoker are being counted in cancer risk. we're only interested in people who choose to smoke. public smoking bans make secondhand smoke less risky/relevant as a factor. we're only interested in the risk , independent of secondhand smoke, of someone choosing to smoke getting cancer.

[margalabargala]

> we're only interested in the risk , independent of secondhand smoke, of someone choosing to smoke getting cancer.

No, that's where you're wrong.

You are only interested in that independent risk.

I, and many others, are interested in how much smoking changes that risk.

Picking random numbers, let's say smoking gives you a 10% chance of lung cancer. It's fine for you to only care about that 10% number, you get to care about what you want to.

But for the rest of us, when making informed decisions based on risk, it matters whether smoking changes it from 9.9% to 10%, or 0.1% to 10%.