How to Generate Bogus Conclusions (E-Cig Study Edition)

[DanTex]

121. But your argument as to why this study is "bogus" is invalid.

You were assuming that the study was uncontrolled and apples-to-oranges, which is not the case. You also assumed that the study didn't take into account desire to quit, which is also not true. As far as the sample size, there are well-understood formulas that govern the relationship between sample size, number of variables, and confidence intervals. I'm not sure why you think that the authors of this paper aren't aware of or capable of using these formulas.

And by the way, a study with 949 people, with only 88 in one group, is not equivalent to a study of 2*88=176 subjects, as you imply in the OP. This is particularly true of the multivariate controlling, because the entire group of 949 will be used to compute the effects of the control variables.

I agree that the conclusion "e-cigs are not helpful for quitting smoking" is overclaiming this study. For one, the fact that a study doesn't find an effect doesn't mean there isn't an effect -- in fact, at a theoretical level, it is impossible to prove that there is no effect, all you can do is fail to prove that there is an effect. Moreover, I would like to see a clinical trial (or, preferably, many of them) before drawing any such conclusion. Clinical trials don't have the same causality problems as observational studies, and, unlike the link between smoking and lung cancer, where a clinical trial would be not just very difficult, but also hugely unethical, it is pretty easy to set one up for e-cigs.

My guess is that e-cigs will prove helpful in that setting. Common sense dictates that they should work at least as well as patches or gum, and probably better, because it simulates the effect of smoking. Still, it is interesting that this study found no effect. I would also have guessed that e-cig usage was associated with a greater likelihood of quitting smoking. But, to the point of the OP, this isn't some dishonest bogus study. Yes, the press release overclaims the results, but that's pretty common, unfortunately.