Lies and Statistics: A Mathematician Reflects on Moral Leadership (2005) - Mary Wheat Gray
"We do science, we don't do policy," I heard from the Food and Drug Administration risk assessors at a recent conference on variant Creutzfeldt-Jakob disease. To me this was an eerie variant of the "I was just following orders" defense offered last year by participants in the Abu Ghraib prison scandal. Adherence to the codes of the Geneva Conventions was considered to be optional by the military guards in the latter case, if indeed they had ever heard of the Geneva Conventions. But such internationally agreed upon standards of conduct are not optional. Indeed, they are mandatory, and it is up to those in positions of authority — leadership positions — to make this clear, in both word and deed.
Out of the horrors of World War II came not only the Geneva Conventions but another important code of conduct, one addressing specifically the obligations, not of military personnel in combat situations, but of scientists engaged in medical research. Among other provisions, the Nuremburg Code requires that:
The voluntary consent of the human subject is absolutely essential.
The experiment should be such as to yield fruitful results for the good of society, unprocurable by other methods of means of study, and not random and unnecessary in nature.
No experiment should be conducted where there is an a priori reason to believe that death or disabling injury will occur.
The degree of risk to be taken should never exceed that determined by the humanitarian importance of the problem to be solved by the experiment.1
Writing of his experiences at Auschwitz, Primo Levi said: "Monsters exist. But they are too few in numbers to be truly dangerous. More dangerous are? the functionaries ready to believe and to act without asking questions."2 Equally dangerous, I would add, are scientists who do not inquire about, or seek to influence, how their work is eventually used. Not to consider the downstream risks involved in one's research is a moral lapse. This was what the Nuremburg Code was trying to prevent in response to atrocities committed against Jews and other "undesirables" detained in German concentration camps during World War II. It is not entirely evident today that it was successful in doing so.
When we think of the moral responsibilities of scientists, we usually think of those whose work has a fairly direct destructive effect: for example, the men who conducted the infamous Tuskegee syphilis experiment, the German physicians working in Nazi concentration camps, or their successors who participated in the tortures during the Argentine and Chilean "dirty wars" of the 1960s. But there are others who often work behind the scenes — statisticians, for instance — who also share responsibility for the outcomes of the experiments they perform. For example, insufficient care in the design, implementation, and follow-up of clinical trials of various drugs may result in unnecessary deaths or illnesses. So the work of a statistician, who must determine these kinds of probabilities and possible outcomes, is very significant. Crunching numbers, scrutinizing data, and arriving at potential outcomes all have an extremely important, though often less obvious, effect on how public policies are established. So the statistician working behind the scenes needs to be as concerned about making moral decisions as the physician, the business person, or the politician who is more often in the public eye.
There has been a lot of talk recently about morality and politics. In spite of the rhetoric to which we are subjected, recent public policy debates have really evidenced little in the way of moral leadership. Consider, for example, the issue of Social Security. Concerted efforts have been made over the last year or so to terrify those under the age of fifty with the scenario that the "safety net" of benefits will no longer be there when they are ready to retire. This is inevitable, we are told, unless radical action is taken, and quickly. Symptomatic of the dubious morality on display is the introduction of private savings accounts as a viable solution to the problem. The very self-centered justification for these is the oft-heard statement, "It's your money, you worked for it, you should get it back."
Leaving aside the fact that even the most enthusiastic proponents of individual retirement accounts must admit that this "solution" will not solve the financial problem at hand, preserving one's savings was never the fundamental principle underlying the Social Security insurance initiative. When the program was established over seventy years ago, the idea was that, collectively, people would be provided for from money contributed by all participants in the plan. This is really no different from government funding of fire, police and defense protection — not to mention public education — regardless of any immediate need for these services. Social security was the lynchpin of reforms made in the 1930s, changes that included regulation of banks, financial markets and conditions of labor. All of these were designed to be programs of government protection for those less able to care for and protect themselves.
So where is the moral leadership from our politicians on this issue today? Research has shown that race- and gender-differentiated life expectancies are being cynically exploited by many of our representatives in government. In a particularly reprehensible attempt to generate African American support for Social Security reform, it has been suggested that the current system actually discriminates against them. The argument is deceptively simple: because their life expectancy is so much less than that of white Americans, most can expect not to receive their "fair share" of retirement benefits. But on closer analysis, this line of reasoning — like so much of what we hear in the public arena — does not stand up to statistical scrutiny. After the age of sixty-five — i.e., the age at which men and women in this country are eligible for Social Security benefits — life expectancies of whites and African Americans differ very little. The racial difference in life expectancy at birth can be traced to early deaths that result, at least in part, from the effects of inadequate public health and welfare policies. Because African Americans have always been disproportionately represented in the low income range, they are actually the ones who are most likely to benefit from the safety-net features of a system like Social Security. The numbers don't lie, and they can easily expose the manipulative double-speak coming from so many of our government officials.
Consider another life-expectancy-based reform proposal that has surfaced in the Social Security debate. Women, the proposition goes, live longer than men, and thus should receive lower monthly benefits in order to "equalize" lifetime payout (and cut costs). On the average, women do live longer than men, in part because of less involvement in hazardous activities and risky lifestyles, but this says nothing about the retired woman who must pay her monthly bills with current income, not with "lifetime expected" income. Moreover, a statistical argument made to the Supreme Court in 1986 established that employment-related pension benefits must be sex-neutral, because the death ages of eighty-six percent of men and women alive at age sixty-five can indeed be matched, leaving only seven percent of the population at either end unmatched. Even for those women whose death ages are not paired with those of men, it distinctly does not mean that they live longer than all men, as our officials in Washington seem to be suggesting.
Nearly twenty years ago, I joined with others to make the case in the courts, and in Congress and state legislatures, for gender equality in the work place; I would hope that the current Social Security controversy does not require that the issue be raised a second time. But, if necessary, I will find myself once again preparing Congressional testimony, using statistics to engage in responsible public policy debate.
The morality of the use of life expectancies to determine benefits is not the only question that involves statistics. Probabilities are involved in other kinds of insurance programs, and these are largely unregulated by state or federal law. For example, in most states young male drivers pay much more for car insurance than young female drivers, on the grounds that they are more likely to have accidents. In some places, a young male driver, with a clean record, may pay more than an older driver — male or female — with two DUI convictions. This is a dubious use of probabilities. Or consider a similar situation involving medical insurance. Those needing psychiatric help frequently forego the assistance, fearing increased rates or insurance denials, not to mention adverse employment decisions allegedly made on the basis of probability. With the relatively recent rise in genetic research, we have also seen a heightened concern as to whether conditions clearly beyond one's control should, based on statistical evidence, affect many aspects of one's daily life — from reproductive decisions, to access to medical care, to employment opportunities. What is the responsibility of the statistician? Is it to manipulate the numbers for the interest of profit, or to ask the right questions in the hope of arriving at what is factual? More important, should we not be concerned about how our research will ultimately be utilized?
Morality should not be passive, i.e., merely refraining from harmful or unacceptable behavior. Each of us has an obligation to be proactive, to exercise leadership, regardless of our profession. This is especially true for those who teach. It is neither sufficient nor desirable to teach ethics or morality as an isolated topic; rather, it should be integrated into all of the disciplines. Similarly, moral responsibility should infuse all of our teaching. How we instruct our students in science and in other fields will be extremely influential in how they perform in their professions, and how they conduct their personal and private lives. In short, we should take the lead in discussions of our students' personal vocations.
The Hippocratic Oath tells physicians to "do no harm." But this ideal — merely refraining from harmful action — is not enough. A recent ethics code proposed in Science, the publication of the American Association for the Advancement of Science, seeks to hold scientists responsible for the purposes for which their research is ultimately utilized. Even though they may not have total control, scientists are obligated to strive to prevent unethical use of their work, and to refuse to participate in projects that have reasonably foreseeable undesirable consequences. In addition to this, the article goes on to say, scientists should seek to uncover and expose the unethical conduct of others in their field. I would add to this a less invasive moral imperative: we must be active models for others — students, colleagues, friends and acquaintances — and we should not shy away from including discussions of ethical principles in our teaching and mentoring.
In addition to my work as a statistician, I am also an attorney, and thus I feel I have an obligation to do pro bono work for the public good. I believe that each person should accept this responsibility, and particularly those who, like most scientists, have received public support for all or part of their professional training. It is easy to see how a statistician can be of assistance in interpreting such evidence as DNA, fingerprints, bullet lead, and the occurrence of Sudden Infant Death Syndrome, but often I am asked why I, as a mathematician, would choose to work on human rights causes in the U.S. and abroad. This is really not as unusual as it may seem. Indeed, for decades the American Mathematical Society has had one of the earliest and most effective human rights committees among all the professional organizations; mathematicians have had a long, though often inconspicuous, history of accepting their responsibility as moral leaders. And our concerns are both national and international. In my first experience of hands-on human rights work, I joined Brazilian, French and Mexican mathematicians in Uruguay to help secure the release of one of our colleagues from prison. There really is an international brotherhood of scientists — with too few sisters, I should add — and it is often very quick to meet its obligation to its members. As in many such efforts, success ultimately depends on the vigorous moral leadership of a few who agitate among their colleagues to gain support for campaigns against governments around the world who actively suppress human rights.
There can be great personal fulfillment in trying to do the right thing. Because of my commitment to using my training to help those who are less fortunate, my life is far more interesting than I ever could have expected. I have been able to combine my legal and statistical expertise to work on human rights and education issues — whether in the Philippines, or Nicaragua, Palestine, Brazil, Iraq, Tanzania, or many points in between — and this has been enormously gratifying.
I was privileged to grow up in an atmosphere where there was no obvious discrimination. I now realize that this may have been due to the fact that the few Jewish, African American and Native American families living in Hastings, Nebraska, at the time were not numerous enough to constitute a perceived threat to anyone. My childhood and adolescence were pretty well shielded from reality. I knew few for whom daily living presented a real challenge.
I can trace my earliest awareness of discrimination, and my resulting activism, back to my graduate school days. At that time, the supermarkets in Lawrence, Kansas, refused to hire "carry-out boys" from the local Native American boarding school. A demonstration consisting of filling grocery carts with expensive frozen items and abandoning them at the check-out counter convinced the store owners that a change in policy was required. Strangely, the notion that there should be "carry-out girls" seemed never to have occurred to us. One would think that such obvious occupational segregation has disappeared, but just a few nights ago I took up with the manager of the Kennedy Center restaurant in Washington, D.C., the question of why all the waiters there were male. Yes, great strides have been made in the last few decades, but more still needs to be done.
My initiation to discrimination on the basis of gender was very personal: one faculty member in graduate school criticized me for taking up a place in his classroom that a man might have otherwise had, and a dean later opined that women do not get Ph.D.s in the sciences because they "just aren't that kind of people." Since no one at Hastings College ever suggested to me that my mathematics and physics majors were in any way extraordinary, these comments came as quite a shock. Shortly after I received my graduate degree I helped found the Association for Women in Mathematics, an organization that has done much for women and girls interested in a discipline that is commonly perceived as being exclusively for men. But what a disappointment recently to hear once again — and this time from the President of Harvard University — that women perhaps do not have the innate ability to do science, or to manage the workload required in the field. What an example of the failure of moral leadership, not only because of his attempt to discount the effects of the discrimination that still exist, but because he chose to pontificate on an area of research about which he apparently knows very little. Those of us in higher education have an obligation to provide leadership through example on the ethical conduct, presentation and discussion of research. We must also demonstrate a genuine respect for human rights, including the right to be free from discrimination.
For several years I worked as a volunteer for Amnesty International and had the unfortunate opportunity of observing widespread violations of human rights and severe suffering in many countries of the world. One of my most disappointing experiences, however, was when, as chair of Amnesty International USA, I visited with governors and attorneys general from around the United States to try to get them to commute the death sentences of those about to be executed. Well-documented research has shown that vengeance motivates most death penalty supporters, but in the case of these officials the motivation was almost exclusively political. "The voters support it!" they said. Perhaps more honestly they might have added, "...and I want to be re-elected." Rarely did I see any deep belief that the death penalty was an appropriate punishment or that it would be a deterrent for criminals. This is an area that cries out for moral leadership, not blind submission to perceived public opinion. What is popular in the minds of most Americans will usually not be what is morally acceptable, and what is morally acceptable will not often be popular. But what are our options when there appears to be little chance of altering public policy?
In the days leading up to the Iraq war, I was opposed to the unilateralism — some would say imperialism — of U.S. foreign policy. Many of my students vehemently agreed with my stance, while others were as adamant in their disagreement. Being in Washington, few had no opinion. Fortunately, my university has a reputation for accommodating, indeed protecting, a diversity of ideas on any issue. As the war progressed, there were many opportunities on campus for opposing views to be aired. In situations like these, I asked myself, what responsibility does a faculty member have for providing leadership? Certainly continued protest on my part was an option, but seeing the hardships endured by the Iraqi civilian population, one could argue that something more needed to be done, no matter how minimal the effect. I decided to go to Iraq to work on a project for the schools, a small effort, but one that I hoped would help. Trying to counteract the bad effects of American public policy through personal involvement is a mechanism I have used before when confronted with government procedures that I found morally unacceptable — for example, lecturing at a Nicaraguan university on time statistical methods to be used in weather forecasting while the U.S. government was supporting Contra attempts to overthrow that country's government. Putting a human face in the place of impersonal and oppressive government policies, whether in Nicaragua or Iraq, at least demonstrates that there are some American citizens who are vitally concerned with the human tragedy involved in the situation.
Engaging in positive activity, whether to ameliorate the effects of morally unacceptable public policy or to ensure that one's work is not used for unacceptable outcomes, is one aspect of moral leadership. Adherence to an ethical code of conduct must also pervade one's professional and personal life, and in this being proactive is preferable, and more effective, than being passive. Perhaps most important to keep in mind, leading by example is far better than seeking to direct others through exhortation. There is enough of the latter in our world today: "Do as I say, not as I do." But hope accompanies the rise of every new generation. Moral leadership for the new millennium must have its source in who we perceive ourselves to be — that is, human beings with the capacity to empathize with others — and must then inform our every deed. For those of us in the sciences, this will involve the recognition that our research has ramifications for our communities, for our country, and for our world, long after it has left our desks or laboratories. Awareness of this profound interdependence of our work and our world must be the place from which we begin and the place where we must necessarily end, lest Primo Levi's statement about complacent human beings — those "ready to believe and to act without asking questions" — become less an observation of the past than a prophecy for the future.