Yes. First of all, on the study in the tobacco control issue of 2016, my colleague and I did an extensive analysis of those papers, and the Cancer Council Victoria, which undertook the original research, responded with a press release saying that the survey was “quite explicitly not designed to assess quitting success or change in smoking prevalence but rather focussed on the immediate impact of the legislation...”. So all of those studies on tobacco control don't do what you just quoted them to have done, and the authors of the studies actually said that.
Turning to the PIR—which is a very, very impressive econometric technique that was undertaken—it found that there is a 0.55% decline in smoking prevalence as a result of the plain packaging policy. What the PIR did not report was that the sample error in their study was bigger than the policy effect size they found.
The other thing that is not clear from the study is that the smoker they built their model on was an unmarried Australian-born 14- to 17-year-old male with a tertiary qualification, employed full-time, but with an income of less than $6,000, and living in Victoria. Now, no such person exists, so it is unsurprising, when you model whether a person who does not exist gave up smoking, and your effect is smaller than the sample error in your data, that you would want to keep that a bit quiet.
The other thing is that the pseudo-R squareds were less than 10%, so while the analysis was very clever, it excluded price. It's entirely, utterly unconvincing.