Michael Wolf
This page was last edited on at
-
Michael Wolf, Tobacco Tactics, updated 03 February 2020, accessed 16 April 2024.
Michael Wolf is Professor of Statistics and Applied Econometrics at the University of Zürich, Switzerland.1 2
On his curriculum vitae, Wolf lists his consulting experience, including consultancy for Philip Morris International (PMI).
Philip Morris International Commissioned Expert
Wolf and his colleague Ashok Kaul, a Professor Economics at Saarland University, Germany, have produced two working papers on behalf of PMI the first in March 20143 and the second three months later in June.4
In both studies Kaul and Wolf conducted a statistical trend (time-series) analysis of smoking prevalence rates in Australia between January 2001 and December 2013. Their first study looked exclusively at 14-17 year olds and their second study included all smokers over 14 years. The objective was to determine whether there was any evidence to suggest that plain packaging introduced in Australia in December 2012 had any impact on smoking prevalence 13 months post-implementation. The authors concluded in both studies that there was no evidence that plain packaging policy had reduced smoking prevalence in the year following implementation.
Talking about the 14-17 year olds study Wolf stated:
“…what we see every single month is that the interval is both below and above, meaning that the lower end point is negative and the upper end point is positive, so the zero is always in the confidence interval, meaning there is no statistical significance for any effect, be it positive or negative.” Michael Wolf, p11, giving evidence to the Chantler review committee, March 20141
He also stated:
“Obviously, a very, very small effect will never be picked up. You can always argue that if you say there is no evidence but maybe there was a tiny, tiny effect.”1
The following page provides details of the considerable critique, from academics and public health professionals alike, of Kaul and Wolf’s two 2014 working papers.
Did PMI Have Any Influence On the Studies?
In early 2014, Sir Cyril Chantler undertook a British-government commissioned review of the evidence for plain packaging policy. In March 2014, when giving evidence to the review committee, Kaul stated that PMI had made first contact with the researchers and confirmed that the tobacco company commissioned the research:
“So as you may know, this work was funded by Philip Morris International…The way Philip Morris approached us was: I was working for a big German consultancy as a technical adviser on a paper on the effects – on the labour market effects essentially – of the new TPD, new Tobacco Products Directive, in the EU so they got to know me, that’s why they approached me…”1
Kaul added:
“Philip Morris International provided the funding for this research. At no time did we provide Philip Morris International with access to the underlying data.”1
What did the contract say?
The Swiss Public health campaign organisation, OxyRomandie, submitted a freedom of information request to obtain from the University of Zürich a copy of its contract with PMI. The contract revealed the following details:
PMI reserved the right to read the papers in advance of publication and suggest changes.
“University agrees that, prior to submission to publisher of a manuscript describing the results for publication, University shall forward to PMIM Morris International Management SA 30 days prior to planned publication a copy of the manuscript to be submitted to PMIM for review and comments and University will take into account in good faith the said comments.”
The contract included a clause allowing PMI to control communication regarding the study, including to the media.
“If at any time either Party or either Party’s Personnel is contacted by a third party, including any news organization, concerning the Services provided under this Agreement, such Party and/or such Party’s Personnel shall make no comment, notify the other Party of the third party contact and refer the third party to such other Party and/or coordinate the information provided to the third party with such other Party.”
The contract was a secret document.
“Neither Party nor its personnel shall, without the prior express approval of the other Party, (i) advertise or otherwise publicize the existence of Terms of this Agreement or any other relationship between the Parties (…)”
Critiques of Study One and Two
Aside from being commissioned by PMI, the report was criticised by public health academics and advocates as providing “neither an unexpected nor a meaningful conclusion.”567
Critiques of the 2014 reports published in The Lancet, Tobacco Control and on the Cancer Council Victoria’s Plain Facts website respectively, argued that:
- One year since the introduction of plain packaging is not necessarily long enough to detect a meaningful effect on smoking prevalence and that this was not the explicit intention of the legislation in the short-term. For example, in their 2015 Tobacco Control article Laverty et al, cited another peer-reviewed research article which collated the opinions of 33 tobacco control experts on the likely impacts of plain packaging on adult and youth smoking prevalence, as well as smoking initiation, two years following its introduction. 89 The experts expected that the largest effect would be on the number of children taking up smoking.
- Given that the aim of the policy was to reduce smoking uptake in young people, smoking prevalence was not the appropriate measure to evaluate the policy. Diethelm and McKee stated in their 2014 critique of Kaul and Wolf’s working papers, published in Tobacco Control, “Measures of ‘prevalence’ cannot distinguish those who took up smoking after plain packaging was introduced from those who took it up previously, with the former likely to comprise only between one-quarter and one-third of the age group.”6
- The small monthly sample in Kaul and Wolf’s studies prohibited any credible analysis of change over a short period of time;
- The study gave no details explaining whether or not the sample of 14-17 year olds was representative of the 14-17 year old Australian population in general;
- Kaul and Wolf were “opaque” when discussing the statistical power of their analyses. See page 8 of OxyRomandie’s Errors and issues with Kaul and Wolf’s two working papers on tobacco plain packaging in Australia for a full explanation of this critique.
- Australia has adopted a comprehensive approach to tobacco control which has contributed to declining smoking prevalence over many years – for example, large excise tax increases, strong smoke-free laws, large scale mass media campaigns, rotating health warnings, cigarette display bans and so separating the effect of plain packaging from the cumulative denormalising effect of all of these policies together is difficult in the short term;10
In addition to these critiques, a peer-reviewed time-series analysis by OxyRomandie’s Pascal Diethelm and statistician Timothy Farley – using the same data that Kaul and Wolf used in their second working paper – concluded that there had in fact been a 3.7% reduction in smoking prevalence amongst adults in Australia following the introduction of plain packaging (even after adjusting for the effects of other tobacco control policies in the country).7
Complaints Sent to University of Zürich
- On 29 January 2015 OxyRomandie sent a letter to the University of Zürich requesting that both papers be retracted and attaching a critique of both of Kaul and Wolf’s studies.
- On 16 February 2015, Kaul and Wolf issued a press release to “ask the anti-smoking organization OxyRomandie and its president Mr. Diethelm to stop their defamatory campaign against us and the University of Zürich.”
You can read Kaul and Wolf’s full response to OxyRomandie’s critique here. In March 2015 an independent re-analysis of Kaul and Wolf’s work by Professor Ben Jann (commissioned by the University of Zürich) was published. 11 Jann is a Professor in the Institute of Sociology at the Univeristy of Bern, Switzerland. 12
- On 14 December 2015, OxyRomandie sent another letter to the University of Zürich which included a 25 page response to Jann’s re-analysis. OxyRomandie maintained that Kaul and Wolf’s work had ‘crucial defects’. The letter asked the University of Zürich to publish OxyRomandie’s critique and the peer-reviewed Diethelm-Farley Tobacco Prevention and Cessation journal article alongside Kaul and Wolf’s working papers on the University’s website. Furthermore, OxyRomandie asked the University to add an additional note alongside Kaul and Wolf’s working papers to highlight that they have been discussed controversially and that the note “should also draw the readers’ attention to their defective nature and to the misleading character of their conclusions.”
University of Zurich Commissioned Review of Kaul and Wolf’s 2014 Research
Jann’s reanalysis of the data used in Kaul and Wolf’s working papers was published in March 2015. A few months later, in August 2015, the IPE issued a statement on Jann’s findings. 13 The statement quoted excerpts from Jann’s report (Image 1) and argued that there was no basis for OxyRomandie’s ‘defamatory campaign’. However, a full reading of Jann’s text provides the full context of these statements (Image 2) which includes criticism of Kaul and Wolf’s methodology.
Assumption of Linear Trend
One of the most vehement arguments between the warring factions is over statistical methodology. OxyRomandie have argued that the basis of Kaul and Wolf’s analyses in both 2014 papers suffer from “design misconception” because they assumed that there was a linear decline in smoking prevalence prior to the introduction of plain packaging in Australia and that this linear trend would continue in the absence of further tobacco control interventions.
“the assumption of linearity throughout the 12-year “pre-treatment” period is a fundamental error which invalidates prof. Kaul and Wolf’s findings. It leads to a simplistic and flawed model, which ignores the well documented effect of tobacco control measures on smoking prevalence in Australia and fails to take into account important confounding factors that could mask the effect of plain packaging. Rather than questioning the validity of the model and revising it, prof. Kaul and Wolf have questioned the data, arbitrarily amputating them by cutting off the lower third, to force them to fit their model. We think that this is wrong and a major flaw of their approach. We respectfully observe that prof. Jann may have fallen into the same trap, although his one-step approach based on logistic regression with a treatment indicator variable was a move in the right direction.”
While, it is true that smoking prevalence has decreased over the previous decades in Australia, declines have coincided with tobacco control interventions including a smoking ban in indoor public places (and outdoor eateries in most states), tax increases and continual anti-smoking mass media campaigns and are therefore not linear.15
Jann also suggested that Kaul and Wolf’s assumption of linearity may be incorrect. The following statement is taken from Jann’s report:11
“Kaul and Wolf assume a linear time trend and hence base their analyses on a linear fit to the pre-treatment data. Deviations from the extrapolation of the linear fit into the treatment period are then used to identify the effect of the treatment. The assumption behind such an approach is that the time trend would have continued in the same linear fashion as in the pre-treatment period if there had been no treatment. The problem is that it is hard to find truly convincing arguments for why this should be the case (no such arguments are offered by Kaul and Wolf).”11
Both 2014 Studies Cited by PMI in its Submission to the 2014 UK Consultation on Plain Packaging
PMI cited both Kaul and Wolf’s 2014 working papers at length in its submission to the second UK consultation on plain packaging which took place in 2014.16 PMI acknowledged in its submission that it funded both studies.
“In both studies, using standard techniques for statistical analysis and applying the standard statistical significance level of 5%, the experts found no evidence that “standardised packaging” had had an effect on smoking prevalence among Australians aged 14 to 17 years old (in the case of the March study) or Australians aged 14 and above (in the case of the June study). Kaul and Wolf confirmed that if there had been an effect in reality (including of the magnitude predicted by Pechey and the DH Department of Health), it would have been reflected in the data. According to the study, however, no effect was found.”
Both 2014 Studies Cited in PMI’s Submission to 2015 Plain Packaging Consultation in Norway
Kaul and Wolf’s two working papers were cited by PMI in its submission to the plain packaging Consultation in Norway in 2015.17
Interview with Christopher Snowdon
Christopher Snowdon holds the title of Research Fellow at the Institute of Economic Affairs, which accepts funding from tobacco companies.18
In April 2014, following the publication of the Chantler Review, Snowdon published an interview with Kaul which criticised the review for omitting Kaul and Wolf’s study of smoking prevalence among minors, even after the authors met with the Chantler review committee to discuss their results. On his blog, Velvet Glove, Iron Fist, Snowdon referred to Kaul and Wolf’s work as “the only piece of real world evidence about underage smoking rates in Australia since plain packaging was introduced”.19
In the interview, Kaul rejects the small sample size criticisms of his study and states that “In a nutshell, our approach would detect a non-negligible effect of plain packaging on smoking prevalence of minors in Australia with a pretty high probability – despite the small sample size.”19
Australian Routine Data Shows Decline in Smoking Prevalence
While Kaul and Wolf argued that plain packaging did not decrease smoking prevalence in Australia, findings from the Australian Institute for Health and Welfare (AIHW) survey found that smoking prevalence decreased significantly among daily smokers aged 14 and over from 15.9% in 2010 to 13.3% in 2013. The AIHW survey is conducted every three years.
In the meantime, data from the Australian Secondary School Students’ Alcohol and Drugs survey collected in 2014 were published in November 2015. The survey showed a 4% smoking prevalence in 12-17 year old adolescents – a historical low.20
For continued updates on smoking prevalence in Australia click here.
For more information on other tobacco industry-connected reports evaluating the impact of plain packaging see Plain Packaging in Australia and the Cancer Council Victoria’s Plain Facts website.
TobaccoTactics Resources
Relevant Links
