According to inforesearchlab (2009), a research center, each year around half a
million Americans die of tobacco related diseases, fifty thousand of them die from
secondhand smoke. It is alleged that tobacco consumption has a negative effect on people's
cardiovascular system and it increases stress levels, to say nothing of the fact that smoking
causes a bad smell. With virtually unlimited resources (or at least being perceived as such), it
is both easy and common to portray tobacco companies as a type of Disney villain, guilty of
world destruction and scheming how to make everybody miserable. Thus, tobacco
companies have been accused on countless occasions of denying that nicotine is addictive, of
hiding the fact that smoking can be harmful, of abnegating the link between smoking and
cancer (despite evidence to the contrary). What is perhaps even more serious than their
simple denial of knowing everything about the deleterious effects of tobacco-smoking, or
defending the free-choice argument simply because it is profitable at the given time, is the
fact that the tobacco industry has reduced public trust in medical and scientific studies.
Indeed, many scientific research papers devoted to the connections between smoking and
various ailments that have been commissioned or funded by the tobacco industry are
notorious for their failure to keep up with decent scientific standards and rigor. This leads to
an ambivalent perception of the case by many people, while there is no rationale for it. This
paper will focus on the tactics used by the tobacco industry to deny the link between
deterioration of health and environmental smoking. In particular, it will address the question
of how epidemiological studies were discredited whilst 'alternative' studies were buttressed.
Finally, this paper will contextualize those findings and overview the possible venues for
possible changes in the near future.
Smoking is now one of the major causes of deaths in many countries all over the
globe. Moreover, recent studies (as well as some older ones), have found that not only firsthand
smoke is linked to various health problems, such as lung cancer, heart disease etc.
Indeed, some health problems are also caused by environmental or secondhand smoke which
has been shown to have a negative impact on, for example, the health of smokers' children
(who obviously do not smoke themselves).
The World Health Organization has been trying to publicize this knowledge for a long
time now. According to this organization, at the turn of the century, smoking was the main
cause of over four million deaths annually all over the globe. If the rates of increased
smoking in developing countries continue at their current trend, smoking will be responsible
for a total of over ten million deaths annually. However, as it is said in an article by
Capdevila (2001, p.1) for the Inter Press Service, the efforts of the World Health
Organization have been countered, to a large extent effectively, by the tobacco industry.
These efforts on behalf of the smoking industry have been largely responsible for the lack of
new legislation which would improve citizens' health by reducing the access to tobacco
products.
Epidemiological studies about the possible connections between smoking and health
problems have been around at least since the 1970s. Although public health awareness is
now much greater than at the beginning of the research, the process has been surprisingly
slow. Indeed, there used to be many campaigns funded by the tobacco industry denying the
negative effects of smoking and secondhand smoke on personal health. This is largely due to
the fact that the tobacco industry has long waged campaigns whose aim was to diminish
various studies which had been conducted to prove the harm of smoking. The change was
largely spurred by the rising number of court cases brought by former smokers who had
developed health problems, notably lung cancer, and who accused the tobacco companies of
hiding the information about the link between smoking and those health problems. However,
the tobacco industries continue to play an important role in trying to question the veracity and
reliability of current studies.
The tobacco industry has also attempted to discredit the institutions that are in charge
of researching the area. In a report for the WHO, Zeltner, Kessler, Martiny & Randera (1999)
found that the tobacco industry had staged events to divert the attention from public health
issues, attempted to reduce budgets for the scientific and policy activities, tried to provoke
conflicts between the various agencies that were responsible for studying the situation and
discrediting the institutions involved, notably the WHO.
One of the problems with epidemiological studies, not only with these which are
linked to tobacco consumption but also of more general issues, is their reliance on statistical
methods. And, as everybody knows, "there are lies, big lies, and there are statistics".
Statistical studies, virtually by definition and by epistemic nature, are prone to
methodological problems. Indeed, statistics are used primarily in cases of incomplete
knowledge, particularly about the population. They can infer information about the
population with a certain degree of certainty, but rarely with 100%. The tobacco industry has
organized numerous symposiums during which "independent scientists" challenged methods
and research which would prove that cigarette smoke had negative effects on people's health.
Indeed, in a study by Diethelm and McKee (2006, p. 10-18), the authors demonstrated that
the habit of creating such symposiums could be traced back to 1972. Such symposiums,
financed by the tobacco industry continued for the following decades. Moreover, scientists
who presented the most solid and groundbreaking cases against the tobacco industry were the
ones who would find their methods to be challenged the most severely.
One of the first major statistical problems of epidemiological studies used to
determine the effects of smoke on a person's health stems from the use of multivariate
regression, which is quite problematic, as shown by Cerrito (2009, pp.1-18). The trick with
regression analysis is that it cannot show causal relationships, but it is nevertheless used to
demonstrate them. In fact, regression analysis can only indicate covariance between two or
more variables, but never causality, as shown in Armitage, Berry and Matthews ( p. 645-8).
However, in case of epidemiological studies (and medical studies more generally), it is safe
to assume that the causality goes in one way (one of the first causal problems of regression
analysis being that one does not know if one variable influences that second, or if it is the
other way around, or both influence each other, or if a third influences both), but it is never
certain. This is easy to use in discrediting the analysis by rival studies.
The second problem results from the choice of variables, in particular, control
variables. There is no way to know that the variables which are used in regression are in fact
the ones that explain the response variable's variation, even if they appear to do so (although
there are ways to verify this and increase certainty). One challenge to such a method is to
claim that the control variables are inappropriate because they capture variance that is due to
more important variables, but they themselves are not the reason behind the variance. The
second problem concerns the fact that the tobacco industry can produce (and did produce)
alternative studies which recoded or altered the variables in a seemingly legitimate way (once
again there are no absolute truths) but also in a way that led to the results becoming
insignificant.
Another method involves discrediting the data on which the studies are based. This
can be done in virtually as many ways as a statistician can imagine (while perhaps not being
the epitome of creativity, statisticians can have quite a few rabbits up their sleeves). First and
foremost, one can challenge the assumptions of regression analysis. This would include the
normal distribution of the response variable, whether there is autocorrelation between
variables, whether the errors are random (if this is not the case, this might mean that an
important variable has been left out, in this case the whole study must be redone), whether the
variables are reliable and valid (if they really measure what they are supposed to measure,
and what kind of errors they have) etc... If these are perfectly accounted for (which is almost
impossible), tobacco companies can challenge sampling issues. They can ask if the samples
are representatives of the population. Is the sampling method the most appropriate for this
kind of study? Do they have any bias incorporated into them? Is the sample size sufficiently
large? Is the inference of the information entirely reliable? Whilst being technical, these
details are extremely important to the given study. Moreover, almost none of the questions
can be simply answered "yes" or "no". This strategy was notably used to discredit the
influential study by Takeshi Kirayama, which found that wives of smoking men in Japan
were more prone to lung cancer than those that married non-smokers. Hong and Bero (2002
p. 1415), identify that challenging Hirayama's sample was the major element of the tobacco
companies' strategy to discredit his research.
Thus the main problem of epidemiological studies is that they are both an easy prey
for the tobacco industry and the only meaningful way to prove any links between smoking
and health problems, as shown by Ping (2009). One cannot seriously consider a medical
study that would be based on a qualitative assessment of the problem. Thus, medical
researchers are forced to use statistics, and statistics are based on a rule of thumb and
judgment calls. While the distribution of a variable may seems to be sufficiently normal for
one researcher (and indeed, for 80% of researchers), it is still legitimate to say that the lack of
normality (because nature rarely is a perfect match for mathematical models) may cause
some problems. One researcher will say that non-random errors in the sample should not
significantly influence the results of the analysis, but another one will say that they may still
influence the results to a certain degree, and both would be right (in most cases). However,
for people who do not have the time or will to indulge in technical and often obscure
methodological debates about the validity of one analysis rather than another, will simply
remember: "well there seems to be an effect, but it is not 100% certain, and the results are
inconclusive." Consequently, the tobacco industry has had the perfect weapon against even
the most rigorous epidemiological studies conducted. Considering how easily discrediting
such reports are, it is perhaps more surprising that the tobacco companies spent so much
money on it, rather than the fact that they did try to manipulate the public opinion with nonobjective studies.
The record of the tobacco companies is far from being clean when it comes to the
level of conducting objective empirical studies to demonstrate the effects of smoking (and
environmental smoking) on personal health. Which is more aggravating is that they tend to
promote cigarettes in a manner that would be appealing to adolescents; as a result, their
image becomes even further tarnished. Somewhat ironically this goal of tobacco companies
has also been proven by empirical studies such as the one conducted by Pierce et al (1998, p.
515). However, arguably the most important, and the most negative, effect of the numerous
denial campaigns waged by the tobacco companies is the proliferation of such discrediting
tactics and their acceptance by the general populace.
Manipulation of public opinion with the use of research of dubious origins and
simultaneous discredit of rival research is not something that is new. Recently, a similar
strategy has been employed by the National League of Cities that wanted to discredit the
findings of a study commissioned by the International Association of Fire Fighters and the
International Association of Fire Chiefs (2009, p.2). In this case a study commissioned by
the fire fighter associations proved that the smoke inhaled by fire fighters, in their line of
work, led to increased cancers and pulmonary problems. A rival study was then created by
the National league of cities to discredit the fire fighters' study, questioning the data, the
methodology and the possibility of establishing a casual relationship as a result of the
analysis (a modus operandi virtually identical to that used by the tobacco industry).
Thus, the main problem that arises from the tobacco industry's tactics is the general
disregard for scientific studies (or at least increased disregard) by the general public, as well
as the use of rival studies discrediting the objective research. While challenging the existing
truths is something good in the scientific community, and has led to much progress, the
system of constantly improving knowledge breaks down when it faces non-rigorous, biased
and economically or politically motivated research. This results in lack of faith towards the
methods, and perhaps more importantly the methodology. The tobacco companies have
poisoned the lives of people on more levels than one would perceive at first glance.
