In Brad Rodu’s latest posting, How many Americans smoke?, he examines the discrepancies in various estimates of this figure. His major point is that if we are truly concerned with ameliorating this “single most preventable cause of death and disease” we need good numbers to measure any progress or the lack of it.
From personal discussions I have learned that many people shy away from giving accurate answers on smoking surveys. Other than simple desires for anonymity, this springs from the current cultural perception of smokers. To self identify as a smoker is to admit to being a failed human being. For those who actually admit to smoking, they tend to, just like in any user surveys, reduce their estimates of their smoking. Their figures not only mirror societal disapproval but are also skewed by an amount of smoking that they think is preferable to how much they actually smoke. And as study after study shows, this is often likely a case of falsely believing the wrong amount to be the case rather than actually lying.
Even in those cases of individuals who clearly see their smoking levels, there can be good reasons not to admit to them. I have an acquaintance who, though he is utterly forthcoming with his doctor, will refuse to sign any document that describes his smoking (or any other drug use) because he knows that the company he works for could use this against him at some point. And if not that company, he realizes thanks to efforts of individuals like John Banzhaf the future may find him vulnerable due to a prohibition that takes no other variables into account.
The war on tobacco has resulted in people hiding their behaviors which in turn makes it seem as though those behaviors have declined which gives the impression of a change that somewhat justifies the war; and so it continues.