ГУЛаг Палестины - Лев Гунин

shot up as a result of at least the first French Paradox broadcast:

The study also found that the benefits of wine drinking extended to

people who drank from three to five glasses of wine per day. "What

surprised us most was that wine intake signified much lower mortality

rates," Safer said to the television show's audience.

Overall, the segment should prove a big boost to the argument that wine

drinking in moderation can be a boon to one's health. The segment was

seen by more than 20 million people. "It isn't just information," said

John De Luca, president of California's Wine Institute, "it's the

credibility that comes with Morley Safer interviewing the scientists."

After the first French Paradox episode aired in November 1991 the

consumption of red wine shot up in the United States, and it has yet to

dip.

The Kim Marcus article underlined your failure to question the conclusion that wine

consumption increases life expectancy:

Throughout the episode, Safer didn't challenge the fact that wine is

linked to longer life; rather, he was interested in what it was about

wine that made it unique. "The central question is what is it about

wine, especially red wine, that promotes coronary health," he said.

Safer came to the conclusion that it is not only alcohol but other

unnamed compounds in wine that contributed to higher levels of

beneficial high-density lipoprotein (HDL) cholesterol.

I had already seen that French Paradox broadcast. As a matter of fact, I had watched your

French Paradox story when it was first broadcast on 5Nov95, and even while watching it I

had immediately recognized that your conclusion attributing longer life to wine drinking

was unjustified, and that you were causing harm in passing this conclusion along to a

large audience almost all of whom would accept it as true. At bottom, then, I see

little difference between your French Paradox story of 5Nov95 and your Ugly Face of

Freedom story of 23Oct94 - in each case, you ventured beyond your depth, giving

superficial judgments on topics that you were unqualified to speak on, discussing

questions that your education had given you no grounding in, and causing damage because

your conclusions proved to be false.

In the case of the Ugly Face of Freedom, the number of your errors was large, and the

amount of data that needed to be examined to demonstrate your errors was large as well,

as can be seen by the length of my rebuttal The Ugly Face of 60 Minutes. In the case of

the French Paradox, however, you make only one fundamental error which is to fail to

grasp the difference between experimental and correlational data - and my demonstration

of your error can compactly be contained within the present letter.

The reason that I am able to assert with some confidence that your conclusion that wine

drinking increases longevity is unjustified is as follows. I have a Ph.D. in

experimental psychology from Stanford, I taught in the Department of Psychology at the

University of Western Ontario for eleven years, and my teaching and my interests fell

largely into the areas of statistics, research methodology, and data interpretation.

Everyone with expertise in scientific method will agree with me that your conclusion in

The French Paradox was unwarranted. It is not necessary to read the original research

papers on which you rely to arrive at this same judgment - even the brief review of the

research data in your broadcast, even the briefer review of your broadcast in the Kim

Marcus quotations above - is enough for someone who has studied scientific method to see

that you were wrong. Below is my explanation.

The French Paradox Research

Cannot Have Been Experimental

There are two ways in which data relating wine consumption to longevity could have been

gathered - either in an experiment, or in a correlational study. If the data had been

gathered in an experiment, then it would have been done something like this. A number

of subjects (by which I mean human experimental subjects) would have been randomly

assigned to groups, let us say 11 different groups. The benefit of random assignment is

that it guarantees that the subjects in each group are initially equivalent in every

conceivable respect - equivalent in male-female ratio, in age, in health, in income, in

diet, in smoking, in drug use, and so on. That is the magic of random assignment, and

we cannot pause to discuss it - you will have to take my word for it.

To groups that enjoy pre-treatment equality, the experimenter administers his treatment.

After constituting his random groups, the experimenter would require the subjects in

each group to drink different volumes of wine each day over many years - let us say over

the course of 30 years. Subjects assigned to the zero-glass group would be required to

drink no wine. Subjects assigned to the 1-glass group would be required to drink one

glass of wine each day. Subjects assigned to the 2-glass group would be required to

drink two glasses of wine each day. And so on up to, say, a 10-glass group, which given

that we started with a zero-glass group gives us the 11 groups that I started out

positing that we would need. As the experiment progressed, the number dying in each

group as well as the cause of death, and the health of those still alive, would be

monitored periodically.

There are many ways in which this simplest of all experiments could be refined or

elaborated, but we need not pause to discuss such complications here what I have

outlined above constitutes a simple experiment which in many circumstances would be all

that is required to determine the effect of wine consumption on longevity.

Such an experiment has never been conducted

And so you can see from my outline of what an experiment would be like that such an

experiment could never have been conducted. We know this without doing a review of the

literature, without having read a single paper on wine consumption and health.

Manipulating long-term alcohol consumption in an experiment is impracticable. We know it

because, in the first place, it would be impossible to get experimental subjects to

comply with the particular wine-drinking regimen to which the experimenter had assigned

them. For example, many of the subjects who found themselves in the zero-glass

condition would refuse to pass the next 30 years without drinking a drop of wine. There

is no conceivable inducement within the power of the experimenter to offer that would

tempt these experimental subjects to become teetotallers for what could be the rest of

their lives. The same at the other end of the scale - most people requested to drink

large volumes of wine each day would refuse, and the experimenter would find that he had

no resources available to him by means of which he could win compliance.

And even if the experimenter were able to offer such vast sums of money to his subjects

that every last one of them agreed to comply with the required drinking regimen - and no

experimenter has such resources - then two things would happen: (1) the subjects would

cheat, as by many in the zero-glass group sneaking drinks whenever they could, and many

in the many-glass groups drinking less than was required of them; and (2) subjects who

found their drinking regimens uncomfortable would quit the experiment. Subjects

quitting the experiment constitutes a fatal blow to experimental validity because it

transforms groups that started out randomly constituted (and thus equivalent in every

conceivable respect) into groups that are naturally constituted (and which must be

assumed to be probably different in many conceivable respects) - a conclusion that I

will not pause to explain in detail.

Manipulating long-term alcohol consumption in an experiment is unethical. And we know

that no such experiment has ever been conducted because it would be unethical to conduct

it, and would inevitably lead to the experimenter being sued. That is, it is unethical

in scientific research to transform people's lives in possibly harmful ways. Most

specifically, it is unethical to transform people's lives by inducing them to drink

substantial amounts of alcohol every day for several decades. The potential harm is

readily evident.

For example, drinking 10 glasses of wine per day, or even several glasses, will

predispose a person to accidents. A single experimental subject who consumed several

glasses of wine and then was incapacitated in an automobile accident would be all that

it would take to bring such research to a halt forever. The accident victim might

readily argue that the experiment requiring him to drink wine was responsible for his

accident, and that the experimenter - and the university at which he worked, and the

granting agency that funded his research - were liable for millions of dollars. In

anticipation of no more than the possibility of such a law suit, no granting agency

would fund such research, and no university or research institution would allow it to be

conducted under its roof.

Consuming substantial amounts of alcohol can not only cause accidents, but it can also

ruin health, destroy careers, distort personalities, break up marriages - for which

reason no experiment will ever require subjects to consume substantial amounts of

alcohol over extended periods of time. The possibility of harm, and thus of law suits,

can even be conceived at the low end of the alcohol-consumption continuum. That is, a

subject prohibited from drinking any alcohol might argue that this for him unnatural and

unaccustomed regimen changed his personality, undermined his career, and ruined his

marriage, and with this claim in hand, could readily find a lawyer willing to help him

sue for damages.

And if such an experiment had ever been conducted, it would

be invalid

Manipulating long-term alcohol consumption in an experiment would fail to meet the

double-blind requirement. And although we are certain that an experiment manipulating

alcohol consumption over an extended period has never been conducted, even if it were

conducted, it would nevertheless contain inescapable flaws which would stand in the way

of permitting cause-effect conclusions. For example, you may be aware that the best

experiments are ones that are "double-blind." A "blind" experiment is one in which the

subjects do not know what experimental condition they are in - they might not know, for

example, whether the pill they are swallowing contains a curative drug, or only a

placebo. In our alcohol experiment, they would not know whether the liquid they were

drinking was wine, or only some wine-colored and wine-flavored water that had been

sealed in wine bottles. Already, we see the impossibility of our wine experiment being

even so much as blind. Just about every subject in our wine experiment would

immediately realize what it was that he was drinking. Tinted water is clearly

distinguishable by its appearance and taste and effect from wine. A blind wine

experiment, then, is an utter impossibility. Most subjects would be able to quickly

infer approximately what experimental condition they had been placed into.

A "double-blind" experiment would be one in which neither the subject nor the

experimenter knew what experimental condition any particular subject was in. For

example, the experimenter hands the subject a capsule, but does not himself know until

the experiment is over whether that capsule contains a curative drug or only a placebo.

In our alcohol experiment, a double-blind experiment would involve the experimenter

monitoring the life and health of each subject, but only after the experiment was over

opening up the sealed envelope to find out how much alcohol that subject had been

consuming over the past 30 years. Utterly impossible as well.

The reason that the double-blind requirement is essential is that without it,

confounding factors appear that might be responsible for any observed longevity

effects. For example, subjects aware that they are in a large-alcohol-consumption group

would also tend to realize that such alcohol consumption might harm them, and so they

might attempt to compensate by taking vitamin pills, not smoking, upgrading their diets,

exercising, and so on. Or, they might start eating fats prior to drinking alcohol, in

order to coat their stomachs and slow the absorption of the alcohol. They might do a

large number of things. What is important is that the knowledge of one's experimental

treatment can lead to one or more changes in behavior, and that it is these unintended

changes, and not the wine consumption itself, that could affect longevity, either in one

direction or the other.

Or, here is a particularly plausible confounding that might appear. Imagine that the

experiment attempts to control wine drinking, and no more than that, and that subjects

do faithfully follow the wine regimen that is imposed on them. Nevertheless, the less

wine that they were allowed to drink, the more beer and hard alcohol they would probably

end up drinking, but which would make the initially equal groups unequal on beer and

hard-alcohol consumption. And so then it would be impossible to tell if differences in

longevity should be attributed to differences in wine consumption, or to differences in

beer consumption, or to differences in hard-alcohol consumption.

But while we may choose to pause and speculate as to what confounding variables may

appear, scientific method does not obligate us to do so. We know that confounding

variables are possible in non-double-blind experiments, and the number that we are able

to imagine is limited only by the time that we allocate to trying. If I cared to spend

a few hours thinking about it, I could write several pages of possibilities. If I chose

to spend a few months thinking about it, I could write a book of possibilities. I am

able to imagine confounding variables either improving health or impairing it at the low