Fudging statistics, tobacco control style

Fudging statistics, tobacco control style

There is a whole secondary approach being deployed by those in various sockpuppet tobacco control charities across the pond with our dear American friends, all intent on reducing tobacco prevalence and protecting youth. Their grandiose plan is to *ahem* raise the legal age of sale from 18 to 21. That’s it. They completely ignore the fact that there are age restrictions already in place, albeit terribly enforced by the local authorities (as is the case everywhere else).

This is typically a case of “got to be seen to be doing something” (a common enough problem for those with power and zero clue on how to use it) about the use of tobacco amongst American U18s, without actually doing anything to correct the actual problem which is one of enforcement. So why the buzzy bee in their collective bonnets about raising the age of sale to 21 instead of 18? An internet survey where just over 4,200 U18’s completed a simple questionnaire which has magically become the phrase “x percent of Americans” or “x percent of smokers” all based on a pretty pathetic, small-scale survey which doesn’t even represent a tiny proportion of the US population (which was, as of 2014 approximately 320 Million).

This one survey, similar to the last-ditch YouGov survey on the plain packaging debacle in the UK where the number of those against plain packs suddenly dwindled, is the reason there is a crusade to raise the age of sale from 18 to 21 across the United States. If there was more than one, maybe one from each State it might be slightly more acceptable to base a nationwide campaign on. Admittedly, there are other “studies” they base this campaign on, this one being the latest.

From the study:

Efforts to disrupt tobacco sales to minors through age of sale restrictions can contribute to reductions in youth tobacco use. The objective of this study was to assess attitudes toward raising the minimum tobacco age of sale to 21 years among U.S. adults.

As far as I know, the existing age restrictions on tobacco sales has been in place for a while and hasn’t made a significant impact to tobacco prevalence at all. In fact the data from the CDC suggests that teen cigarette use was 15.8% with the age of sale being 18. Good to know those restrictions work isn’t it? So now the plan is to again increase the age of sale of tobacco from 18 to 21, based solely on a single internet survey.

Tweet on the 70 percent claim

Which of course leads to statements from state funded sockpuppets such as this, who wants to bet that some of the cash revenue from tobacco tax goes to these folks? Not a bet I’m going to take.

So it begs the question, is the “70% of current smokers” being in favor of raising the age of sale from 18 to 21 actually correct?

Frankly, no it isn’t. Which is hardly surprising. It can be interpreted the way it’s stated, but in reality it’s utterly misleading.

Respondents were asked: Do you favor or oppose raising the legal minimum age to purchase all tobacco products from 18 to 21? Responses included: strongly favor, somewhat favor, somewhat oppose, and strongly oppose.

Definition of “Somewhat”

Oh joy another multiple choice question, this time with two supposed “positives” and two supposed “negatives”. The trouble with a “somewhat” answer is the fact that it’s pretty wishy-washy, it’s not a definitive “yes” or “no” answer. It’s like asking someone for a score out of 10, with 10 being “strongly favor” and 1 being “strongly oppose”, it’s a gradual scale and pretty useless in the context of being supportive or not.

If a question such as “do you favor or oppose” a particular proposal, especially in a council meeting or other official and legal environment, the answer must be either favor or oppose, being “on the fence” is abstaining, or have no view.

Data came from Styles, which draws from KnowledgePanel, an online panel recruited using probability-based sampling to reach respondents regardless of landline phone or Internet access.

In 2014, a total of 4,269 participants completed Summer Styles (June–July), yielding a 69% response rate.

As with YouGov here in the UK, Styles selects participants based on certain criteria, either an exact match based on previous surveys or a loose match based on generalised interests in particular subjects. In this case 4,269 participants have either taken surveys on tobacco restrictions or similar subjects in the past and those were “randomised” based on location so you would expect a pretty wide net has been cast across the US for this survey to gain a snapshot insight into the current thinking of the average US citizen that is part of KnowledgePanel or Styles.

Selection bias effect

This is a clear case of selection bias where participants are more likely to take part in the survey because it either directly affects them, or is a subject that they have a strong interest in. You could of course say the same about the Vaping Truth Survey most of the participants of that have an interest in vaping, or are vapers themselves. Sadly, selection bias is part and parcel of most surveys simply because if you don’t have any interest in the subject, you don’t take the survey it is that simple. However, surveys designed for a specific subset of participants can be useful in determining a generalised “feel” on a subject so long as it is clear that the bias exists.

The study based on the survey indicates this in the limitations:

although Styles draws from a panel that has a nationally representative sample, it does not recruit using population-based probability samples.

self-reported smoking status could lead to reporting bias; however, the validity of self-reported smoking is established

Finally, state-level estimates could not be calculated.

Based on that, the survey doesn’t represent the views of the population in general, has flaws with self-reporting and estimates applicable to State level populations couldn’t be calculated. Yet, unsurprisingly the statistics are being bandied about as if they were gospel and representative, which they clearly are not.

Respondents who selected strongly favor or somewhat favor were considered to favor raising the minimum age of sale. Assessed sociodemographics included sex, age, race/ethnicity, education, income, U.S. region, and cigarette smoking status.

Out of the four possible answers, it didn’t matter if you responded with “strongly favor” or “somewhat favor” the answer would be marked down as “in favor”. Need I repeat the definition of “somewhat” again?

The results speak for themselves, somewhat:

Three quarters of adults strongly or somewhat favored raising the minimum age of sale to 21 years, whereas 14.0% and 11.0% indicated they somewhat or strongly opposed it

At least, that’s how the authors interpreted the statistics. They claim, categorically that 75% of the number of respondents are in favor of the proposed age increase but go to great pains to segment the “somewhat opposed” and “strongly opposed” results; which by the way equate to 25%.

Is Tobacco21 correct? Statistically no.

  • Overall : 4,219 Participants
  • Strongly in favor: 50.4%
  • Somewhat in favor: 24.6%
  • Somewhat opposed: 14 %
  • Strongly opposed: 11%

Here’s the kicker, the vast majority in favor (using the author’s definition) are never smokers. The puritanical nanny’s that love to suck the fun out of life.

Breakdown by smoking status

This is where it also gets interesting, and frankly highlights the statistical fudgery that Tobacco21 rely upon to mislead the general public.

  • Never Smokers, 2,254 (55.5% of total participants)
  • Former Smokers, 1,236 (30.3% of total participants)
  • Current Smokers, 585 (14.4% of total participants)

Tobacco21 are claiming that 70% of smokers are in favor of the age increase, but the proportion of current smokers within the survey results is hardly proportionate compared to the 2,254 never-smoker participants. Remember, the majority of never-smokers anywhere in the world have been socially trained by these sockpuppet charities and decades of tobacco control to demonise and stigmatise anyone who smokes.

Also, remember that the “somewhat” answers are statistically meaningless, like putting “None of the above” on a UK Ballot Paper. It’s an answer, but not a particularly meaningful one unless you need to bolster statistics elsewhere.

In truth, it’s only 43.3% of current smokers (256 participants) that, at the time of the survey stated they supported the age increase, whilst the “maybes” (256 in total combining the somewhat favor and somewhat oppose percentiles (43.8%)) with a small (72) percentage (12.4) strongly opposing the proposal, not the “70%” that Tobacco21 are claiming.

By adding the two “in favor” statistics, (256.23 strongly in favor & 152.685 somewhat in favor) you get 408.915 participants “in favor” which is (drumroll please) 69.9% and why not, let’s drop the decimal point and round it up hey?

Just for fun, the percentages based on total participants (regardless of smoking status):

  • Strongly in favor, 2078 (49.25%)
  • Somewhat in favor, 999 (23.69%)
  • Somewhat opposed, 571 (13.54%)
  • Strongly opposed, 428 (10.14%)

Let’s mash-up the “in favor” numbers, just as Tobacco21 would do:

  • In favor, 3077 (72.9%)
  • Opposed, 999 (23.67%)

There you go, looks like my future as a tobacco control statistics fudger is set.

Photo by Leeloo The First