The dark ages of social science.

IN EARLY 2006 five hundred economists electronically signed their names to an open letter to President George W. Bush, trying to convince him that immigration is a net gain for the American economy. But why was this necessary? Was it because American policymakers were ignoring widespread consensus on the issue? Or was it merely the latest propaganda in a debate that threatens to become more about politics than social science?

Can one imagine a group of physicists trying to lobby heads of state in support of their astronomical or physical theory? Actually we can. In the "dark ages" of natural science (which arguably lasted long after the Renaissance liberated European art, music, and literature from the grip of scholastic ideology) it was commonplace for theories of the physical universe to be challenged and rejected based on nothing more than their conflict with sacred religious beliefs. Galileo famously sought audience with the pope, trying to convince him not to reject the new telescopic evidence supporting Copernican theory merely because it conflicted with scripture. Galileo failed and ultimately drew a lifetime prison sentence from the Catholic Church that made him a martyr and marked the beginning of the end for the church's reign over science. Three hundred and fifty years later, Pope John Paul II apologized to Galileo.

While some may grumble that not much has changed, case in point being the Bush administration's head-in-the-sand rejection of the overwhelming scientific consensus about global warming caused by human activity, the outrage that many feel over such ignorance can bolster our commitment to once and for all move past the point where we allow ideological beliefs to trump those based on scientifically gathered evidence. We may not always like what science tells us, but its progress leaves us little choice when trying to reconcile beliefs based on experimental evidence with those based on faith and wishful thinking.

Still, evidence abounds that today we live in a "dark age" for our understanding of the causal basis of human behavior--gut instinct, superstition, and political ideology left and right happily trump the nascent efforts of empirical social scientists to figure out what makes us tick. The question of whether immigration is a net plus for the American economy is a good case in point. Why was it necessary for so many economists to assert so strongly--in a public forum no less--that immigration is good for America? Was it because social scientists are all on the same side of this issue? The scholarly research belies such a conclusion. Cynically, a more likely alternative is that the case in favor of immigration was put so strongly not because the evidence is so overwhelming, but rather because the studies on this issue have remained so equivocal.

Over the last thirty years, numerous studies have shown that immigration is a net plus to the U.S. economy. A nearly equal number have shown that immigration has a net cost, and a handful have concluded that it has no effect at all. But how can this be? Like physical reality, isn't there a truth to the question of the economic impact of immigration? Likewise, isn't there a truth to the plethora of other questions that have vexed social scientists in recent years, like what led to the drop in crime in the United States in the mid-1990s (and why rates are slowly creeping back up again) and whether more stringent gun control might possibly have a crime-enhancing (and not just a crime-inhibiting) effect as it reduces not only the availability of guns to lawbreakers but also presumably many defensive gun uses as well? Yes, these questions all have right and wrong answers, and even though they may be complicated to investigate, the cause-and-effect relationships behind them all seem amenable to study by the empirical methods of social science. Why then have the answers given by social scientists remained such a muddle?

One reason, unfortunately, is that social scientific debate is often fraught with political ideology that influences the outcome of what should be empirical study. Predictably, most of the research that purports to show that immigration is a net cost to the economy comes out of conservative think tanks that are anti-immigration. On the other side of the equation we find a curious alliance between pro-business conservatives and political liberals, who argue that immigration is a net benefit, because they are pro-immigration. In the guns and crime debate we find a similar vetting of social theories through a political lens, where some ideas seem championed primarily because they are in sympathy with fashionable political beliefs, even though they may not square with the data.

My claim here isn't that all social scientists are making up their results any more than that all medical researchers whose work is paid for by drug companies must be fudging their data (though this does occasionally happen). Rather, the claim is that when ideology creeps into empirical research, the results are rightly suspect precisely because there does tend to be such a marvelous coincidence between those who might hope that a given result will be true and those who claim that the data show precisely this outcome.

Any good statistician can tell you the various ways to cheat: rely heavily on confirming instances (and ignore or explain away falsifying ones), use questionable assumptions or proxies for what you are measuring, or simply re-run or narrow your data sets until you get the results that "should" be true. Naturally, there is a stalwart group of methodologically tough-minded social scientists who are outraged at such tactics and who have worked hard to make the social sciences more scientific. But bad social science tends to drive out good and so makes all inquiry into human behavior seem suspect in the minds of those who are skeptical that social inquiry could be rigorous or objective in the first place. Is it any wonder that policy makers have felt free to ignore social scientific work or to cite only those studies that already agree with their ideological positions? But it doesn't have to be this way--the natural sciences have already shown us how to escape from this kind of "dark age" thinking.

Just as in the natural sciences, the way forward is to embrace the "scientific attitude" toward questions that are matters of fact: to be willing not only to hear contrary evidence but to seek it out, even if it clashes with our most closely held religious or political convictions. Although we may care deeply about the results of social scientific inquiry--and may be horrified by what it reveals--it does us no good in the long run to deny the truth about the connection between guns and crime, whether the death penalty deters murder, or what motivates the mind of a suicide bomber, just as it does us no good to pretend that a new drug can cure cancer if it can't. As in natural science, once you understand the causal roots of something, you will be in a better position to explain it and then to use this understanding to solve the problems that have grown up in our ignorance.

Potential examples abound in the newly developing field of behavioral economics, in quantitative political science, and in the last three decades of work in social psychology. For instance, social scientists have known for years that humans regularly violate well-codified norms of rational decision making in our everyday lives. And so after the media reports one or two deaths due to rare side effects of childhood vaccinations, many thousands may choose not to have their children immunized against preventable diseases, thus exposing them to several times the risk of death from polio, rubella, and whooping cough. Even seasoned professionals sometimes violate rational norms, based merely on how information is presented. In one study physicians were asked to choose between two alternative protocols in the face of a disease that was expected to kill six hundred people. In protocol A, two hundred lives would be saved, and in B there was a one-third probability that six hundred lives would be saved. When given this choice, most physicians were "risk averse" and chose protocol A. But when the choice was rephrased so that it was now given in terms of lives lost rather than lives saved, the majority of physicians now chose the "riskier" option, protocol D, where there was a "two-thirds probability that six hundred people will die" as opposed to protocol C, where "four hundred people will die."

Such lapses in reasoning lead us daily to undersave for retirement, drive instead of fly, believe eyewitness testimony, pay higher salaries to better-looking employees, and engage in nuclear brinksmanship, even though such actions fly in the face of what we should know better as the result of decades of work in the social sciences. Meanwhile our politicians stumble their way through issues like crime, welfare, immigration, terrorism, and poverty, heedless of the fact that there are right and wrong answers to the questions behind our social miseries, answers that are accessible through the methods of good empirical social science.

Once the day comes when we fully embrace the scientific attitude about the study of human behavior and accept it as an apt area for experimental and empirical inquiry, it should no longer be necessary to write open letters to the president to try to spin the debate toward a particular outcome. Of course the first step in coming out of a dark age is to admit that we are currently in one. As we face the costs of our ignorance over the last several decades--in cases like Bosnia, Rwanda, Sudan, Congo, 9/11, child prostitution, genocide, and torture--how can we doubt that the primary cause of human misery in the world today is human mistreatment of one another? And yet, after three hundred years of social science, how little do we understand the problems that humans have created, and how much progress have we made in solving them?

The natural sciences have faced such challenges head on. Are the social sciences now up to the same task of struggling against the ingrained political ideologies and self-serving belief that we already basically understand the causal forces that lie behind human behavior, when world events so clearly demonstrate that we do not?

Lee McIntyre is a research fellow at the Center for Philosophy and History of Science at Boston University. He is the author of Dark Ages: The Case for a Science of Human Behavior (MIT Press, 2006).