Inconvenient Facts and Public Policy
Few today would contest the idea that "knowing the facts" is important for policymaking. Public agencies often have units dedicated to collecting and analyzing data, invest significantly in acquiring information from consultants, and routinely try to evaluate the results of their efforts. Not to be outdone, legislatures have developed their own ways of fact-finding: What is now the Government Accountability Office dates to 1921, and the 1946 Employment Act established the Joint Economic Committee, which has no legislative responsibilities but conducts studies and hearings to provide information to Congress about economic and social life. University programs for educating future public employees bristle with instruction in statistics and analytic methods. Unsurprisingly, fact-checking has become a routine part of the media's coverage of government officials; as a 2018 study by the National Opinion Research Center and others showed, 87% of the public — and 99% of journalists — believe the press "should verify the facts." Another 2018 study, from the Pew Research Center, found that 92% of the public believed it was "important" for "people to agree on basic facts even if they disagree politically" (although just 33% thought that this actually happens).
Yet, to paraphrase President John Adams, facts can be troublesome things. They may be widely believed yet possibly wrong. They may offend one or another influential constituency. They may point to actions that are inconsistent with fundamental beliefs — and that are often in conflict with powerful interests. "Everyone is entitled to his own opinion, but not to his own facts" is an observation frequently attributed to Daniel Patrick Moynihan (as well as to former Secretary of Defense James Schlesinger and the financier Bernard Baruch). But what if the "facts" are, as we say today, problematic?
Moynihan was no stranger to problematic facts. In 1965, as an assistant secretary in the Labor Department, he authored a now-famous report on "the Negro family," which identified the increasing share of single-parent households among African-Americans as a serious barrier to racial equality as well as black progress. Although that report informed a well-received address that President Lyndon Johnson delivered at Howard University, civil-rights leaders and others denounced it as "blaming the victim," shifting culpability for continued inequality away from public and private policies and putting it on the shoulders of black Americans themselves. The controversy would dog Moynihan throughout the rest of his long, varied, and illustrious career.
It was also not the last time troublesome facts would create problems for Moynihan. Just four years later, the Harvard Educational Review, perhaps the leading outlet for scholarly research in education at the time, devoted most of an issue to an article by University of California, Berkeley, psychologist Arthur Jensen entitled "How much can we boost IQ and scholastic achievement?" With copious data and complex equations, he reached a simple but explosive conclusion: "[T]he failure of recent compensatory education efforts to produce lasting effects on children's IQ and achievement suggests that the premises on which these efforts have been based should be reexamined." In an introductory note, the Review's editors summarized the crux of Jensen's argument: "[G]enetic factors are much more important than environmental factors in producing IQ differences" among racial and socio-economic groups. Realizing how controversial this conclusion would be, the editors also promised that several prominent scholars would discuss it in the next issue and invited additional comments from readers.
Jensen's article was more than just controversial for Moynihan, a Harvard Education School professor then on a leave of absence to serve as President Richard Nixon's assistant for domestic policy. Moynihan was a liberal and a Democrat who had taken a senior position in what was expected to be a conservative administration. Moynihan swiftly found himself having to deal with a serious piece of research, appearing in a serious journal published by his own institution, a piece that appeared to offer the worst possible reason for reducing or eliminating compensatory education programs: the genetic inferiority of their intended beneficiaries. Moreover, as he later wrote, "Jensen was a liberal Democrat, a respected scientist, a professor at...Berkeley," yet his analysis seemed to give scientific sanction to long-discredited views about race. Reflecting on the episode, Moynihan asked: "What was government to do with this proposition?"
What, in other words, is to be done when the "facts" — as Jensen, at least, saw them — have implications that would be widely regarded as unacceptable and possibly even dangerous in a country that had just endured a year of violent racial conflicts and the assassination of its most prominent civil-rights leader? Moynihan already had a reputation for criticizing the ardent architects of President Johnson's "War on Poverty" for putting their theories of what should be done to help poor minority-group members ahead of a realistic understanding of the challenges these groups faced and what government programs could actually do to address them. That was one of the reasons, it was widely assumed, that Moynihan — a supporter of Robert Kennedy during the 1968 presidential campaign — had been asked to join Nixon's White House staff. Yet in short order, he faced the choice of either rejecting what he had been preaching or accepting a conclusion that might resonate with some of the administration's politically incorrect supporters but would most assuredly be repugnant to almost everyone else.
So what is government supposed to do with this type of inconvenient fact? What is a right-minded policy advisor supposed to encourage it to do?
JUST THE FACTS
The answers to these questions would be easier if the distinction between "facts" and "opinions" were clear-cut. But often it is not. It turns out that facts come in almost as many different guises as opinions do — and the two intersect and interact in more ways than commonly supposed.
Most definitions stipulate that a "fact" is something that is true or real. (Merriam-Webster, for instance, says a fact is "a piece of information presented as having objective reality.") But that assumes some method of ascertaining truth or reality. Any number of instruments can be used to measure temperature, for example. But they may differ in a variety of ways, some relatively small (such as how precise they are) and others of greater significance (such as whether they use Fahrenheit, Celsius, or Kelvin). A thermometer may say it's 32 degrees outside, but how accurate that is and what to make of it depends on its quality and location, whether it's in the sun or shade, or sitting in snow. A reading of 32 degrees Fahrenheit requires a coat, but 32 degrees Celsius would suggest a trip to the beach. Charting trends over decades or centuries, or across continents, will yield copious additional findings, but as the disputes over global warming attest, not more certainty as to what's going on, much less why, much less what, if anything, might or should be done about it.
Where it gets especially tricky and sometimes genuinely complicated is when — as in the case of the Jensen study — a "fact" turns out to be the product of human calculation. Indeed, it could fairly be said that a great deal of what we regard as fact is the product of some sort of reasoning, investigation, or calculation, however elementary. Intelligence, for instance, is usually measured by an "IQ test," the most famous of which are the Stanford-Binet Intelligence Scales, initially developed at the beginning of the 20th century. After considerable research and revision, Stanford-Binet is now viewed by most as a reliable measure of general cognitive ability, or "g." But many have challenged how fairly it measures intelligence among people with different cultural backgrounds, what "IQ scores" actually mean, and whether there is in fact a single "g" rather than multiple types of abilities. A lively scholarly debate is underway today over the extent to which one's intelligence is affected by the environment in which one grows up, and whether it is fixed or mutable.
Although Stanford-Binet continues to be widely used for research and in educational and employment counseling, one can choose other ways of calculating intelligence, too, depending, perhaps, on how the findings they produce line up with one's moral or policy preferences.
Economists depend almost entirely on these "derivative facts." In the United States, Britain, and other developed countries, reliable, standardized systems for measuring national economic data did not begin to be established until well into the 20th century. (The Great Depression was several years along before New Dealers laid down the rudiments of the tools for tallying unemployment that are used today.) A profusion of statistics now exists, purporting to measure such economic facts as national income, productivity, employment, money supply, income equality, and even "happiness." All are produced through complicated calculations and estimations, using information gathered in a variety of ways.
As a result, depending on what information one thinks should be included or omitted, the same economic conditions can be portrayed differently within and between countries, as well as over different time periods, sometimes to confer political benefits (or avoid costs) for governments in power or candidates who seek power. Although two consecutive calendar quarters of decline in gross domestic product are required before a country is technically deemed to be in recession, the beginnings (and endings) are not as clear as this factual definition might suggest. Although the Great Recession started in December 2007, it took the National Bureau of Economic Research — an association of economists that is the official (though not undisputed) arbiter of business cycles in the United States — a year to declare that it had begun, and more than a year after it ended in the middle of 2009 to declare it over.
The poverty line is another example of a derivative fact that owes much to judgment and opinion. In the United States, the modern version was devised in 1963 by Mollie Orshansky, a statistician in the Social Security Administration. It was based on a multiple of the estimated cost at the time of purchasing a "nutritionally adequate" supply of food for families of various sizes. When President Johnson declared his War on Poverty the following year, the Orshansky measure became the official benchmark of its progress.
But this measure is not universally accepted either. Some claimed — and insist today — that it overstated poverty by overlooking in-kind benefits to the needy, such as food stamps and Medicaid, which have grown substantially since the 1960s. Others argued that it understates the number by failing to account for the supposedly higher burdens of work or medical expenses on the poor. Even the concept of a "line" to differentiate those in poverty from those who are not has come into question. After all, some people's incomes put them slightly above, or slightly below, or they fluctuate a bit from year to year. Perhaps a different kind of standard, based on how far a person's income fell from that of the average American, would be better, but that would also mean that some sizable portion of the population would always be classified as poor.
In 2010, a federal inter-agency committee issued a report recommending the creation of a "Supplemental Poverty Measure" (SPM). It proposed keeping a poverty threshold based on the cost of purchasing necessities, but adding in-kind benefits to income while subtracting work and medical expenses along with tax payments. As a result, the U.S. Census Bureau now reports two sets of derivative facts about poverty: the "official measure" and the SPM. The differences have not been large. In 2017, for example, 12.3% of Americans were below the current version of the Orshansky line, while 13.9% did not reach the SPM threshold. But which of these (or any of the other measures still vying for acceptance) should be considered definitive — and therefore affect the distribution of today's enormous amounts of assistance to people who are poor — depends on one's opinion about the meaning of poverty. And that doesn't begin to sketch the dilemmas faced by policymakers asked to consider, for instance, whether preschool should be extended at no cost, partial cost, or full cost to families under 300% of the poverty line.
THE FACTS OF POSTMODERN LIFE
There is yet a third type of fact that has become more prominent lately and is even more difficult to specify. One might call it a "postmodern fact," because it ultimately rests on subjective judgments about real characteristics. For example, when Jensen published his article about racial differences in IQ, few would have questioned the validity of grouping people by ancestry or skin color. But today, one's race reflects not just one's inheritance, as Jensen's research implied, but one's identity, forged through a variety of personal and social experiences. Race and ethnicity are, to use current jargon, "social constructs" at least as much as genetic ones.
The decennial census provides a good illustration of how the facts of race and ethnicity have been changing. The first effort to enumerate the American population, in 1790, needed only three race-related categories: free whites, all other free people, and slaves. A century later, Native Americans and Asians — with different categories for Chinese and Japanese — had been added, along with groupings for people who reported themselves as partially Black. The 20th century saw an explosion in categories, with multiple varieties of Native-American, Asian, and after 1930, Latin-American identities to choose from, as well as an undesignated group, "Other." The 2010 census form listed 18 designations (plus "Other"), and respondents could choose two or more if they were so inclined. While new genealogical tools have made identifying one's ancestry more scientific (as Massachusetts senator Elizabeth Warren discovered, to her embarrassment, when genetic testing revealed that her claim to be a Native American rested on a single ancestor many generations removed), they are also adding much more nuance to what had once seemed to be distinctions that allowed little room for interpretation.
Something similar may be happening in relation to gender. In the beginning, the Book of Genesis records, God created man and woman, but today, amid changing (and often controversial) attitudes toward sexuality, social-media platforms such as Facebook and Instagram allow their users to choose among several dozen categories of gender. Sometimes reluctantly, sometimes not, public agencies have begun to accommodate themselves to this new reality. The 2020 census, for example, will include a question aimed at distinguishing between "same-sex" and "opposite-sex" couples. A bill introduced in 2018, the Census Equality Act, would go even further, requiring the Census Bureau to ask about sexual orientation and gender identity in its next enumeration. For athletics, military service, employment, education, and other areas as well, what had once seemed a fact of life now looks like a choice and, moreover, a mutable choice that can be changed over a person's lifetime.
So facts are indeed troublesome, but not just because, as Adams said, they can be "stubborn" and unalterable by "our wishes, our inclinations, or the dictates of our passions." To the contrary, many facts are troublesome because they may be maddeningly malleable, and the more complex the reality they are trying to describe, the harder it may be to find agreement about them. To one degree or another, our opinions inevitably color our understanding of the facts. And influencing how people come to view the facts is often the first step in affecting their preferences for public policies.
FROM RESEARCH TO POLICY
Yet troublesome as facts are, opinions — the inferences one draws from a particular set of facts (or from invented facts, or no facts at all) — might be expected to cause even more problems. Indeed, if everyone is entitled to his own opinions, there should be no end to them and great difficulty in reconciling them.
Students of democratic government once worried that such a system would be, as the 19th-century historian Thomas Babbington Macaulay put it, "all sail and no anchor," lurching this way and that in gusts of opinion, but never settling on a steady course. More recently, others have forecast government deadlock due to the conflicting opinions held by influential groups. In the United States and elsewhere, more authoritarian systems — constraining, if not suppressing outright, the expression of opinions — have attracted support by claiming to offer, in Johns Hopkins professor Yascha Mounk's phrase, another "gateway to prosperity." (Authoritarian governments can worry less about facts, because they are not inconvenienced by voters who might require them to justify their actions.) On a smaller scale, a number of colleges and universities are experimenting with such a system; in the name of preventing offense to one or another "vulnerable" group, institutions once committed to the free flow of ideas have taken steps to restrict the expression of unpopular opinions, and even the discussion of inconvenient facts. This includes the suppression of academic work that is done with the best of intentions: If it doesn't yield the right results, it isn't welcome.
Social science, broadly speaking, does not enjoy a platinum reputation for reliability, and there are plenty of fields within it where the reputation is downright tarnished. It is no mystery why; the research is sometimes truly dubious, particularly when it's dubbed "qualitative" or "case study." (This is awfully common in education-policy studies, for instance, where it's often easy to locate research "findings" that may be cited in support of every imaginable policy whim or ideological preference.) Other times, however, it's based on the careful, honest analysis and re-analysis of immense data sets. Even when the research is impeccable, however, there's no escaping the crucial insight made by Louis Wirth back in 1936: Every finding touches on somebody's interests. Everyone may be entitled to an opinion, but opinions that claim to be scientific have implications — and possibly consequences — that advance or impede the interests of real people.
Jensen's research, for instance, was controversial not just because it asserted facts about the IQs and educational records of racial groups, but because it sought to connect them. Using methods of social and psychological research, he drew conclusions that were immensely troubling in a country that was then in the midst of a historic effort to improve the lives of African-Americans. Moreover, neither Jensen's credibility as a scholar nor his liberal sympathies could easily be questioned. Many would ask whether he should have been "entitled" to express such a conclusion, even if he believed it, when it was potentially so damaging to some of his fellow citizens.
As inconvenient as Jensen's research was, it pales in comparison to On Equality of Educational Opportunity, a 1966 study whose principal author was James Coleman. This massive statistical analysis has challenged conventional wisdom about education policy ever since. Authorized by the Civil Rights Act of 1964 and authored by one of the nation's foremost mathematical sociologists, it was expected to document widespread racial disparities in access to the resources thought necessary for educational success, such as well-equipped schools and well-trained teachers, and would prove that these differences explained why African-American children lagged behind white youngsters in education. But after surveying over 600,000 students throughout the United States, Coleman and colleagues concluded that the principal disparities were regional, not racial. Moreover, using sophisticated analytical methods, they found that what schools could provide their students by way of resources counted for much less toward educational success than what the students received from their parents and peers. In other words, increasing federal spending on public schools was not likely to do much to equalize educational achievement between white and African-American students.
So unsettling were these findings for a government that had — just one year earlier — enacted the first major federal aid-to-education program that it released Coleman's report over the Fourth of July weekend, a time when the attention of most Americans was likely to be directed elsewhere. But the import of the Coleman Report, as it quickly became known, was not lost on educators and policymakers. Many challenged both the facts on which it was based and the methods leading to the authors' inferences.
Then still at Harvard, Moynihan, together with the eminent statistician Frederick Mosteller, organized a faculty seminar to re-analyze Coleman's data. The group included many scholars who were skeptical of the original report's findings. After taking the criticisms into account, however, the group concluded that Coleman and his colleagues had essentially been correct: Improving the school resources available to African-American students would not bring them much closer to their white peers in educational achievement so long as large disparities in family and community life persisted. Other work growing out of the seminar extended the report's conclusion to include gaps in lifetime income, which also seemed rooted in factors over which government programs had little influence.
That, however, was not the end of it. In the decades that followed, researchers and education experts have repeatedly challenged (or ignored) the Coleman findings. Legislators and courts have continued to make decisions about education policies such as school funding without showing much regard for the questions the report raised about the limited impact that spending has on school success. Large achievement gaps remain.
Much of the effort devoted to policy research assumes that knowledge is a way of reducing areas of disagreement. But the Coleman saga suggests that this is not necessarily so. Like facts, opinions can be stubborn and troublesome, especially when they call into question widely held beliefs, predilections, and philosophies. And while everyone may be entitled to an opinion, some may be more entitled than others — and not just because they rely on better research. Experience, trust, and even wishful thinking may confer legitimacy on opinions despite research findings to the contrary; good research does not inevitably drive out bad. Instead, the existence of competing opinions — especially when expressed in the language of experts — is as likely to give protection to divergent (and sometimes selfish if not objectionable) interests as to undermine them.
The Coleman saga also illustrates how the very same facts — and same conclusions derived from those facts — can point to divergent implications and policy proposals, depending on whose lens is doing the refracting. Coleman himself initially thought his findings made a case for integrating schools so that minority children would have opportunities to attend classes with students from different backgrounds. Other educators believed that they highlighted the need for more effective schools, which would focus their resources on those elements — such as better-trained teachers — that seemed to have at least some impact on how well students from disadvantaged backgrounds fared. Yet another group of policy analysts saw the Coleman report and subsequent research as emphasizing the need to expand the range of options for parents and their children by developing school-choice programs.
During his time as Nixon's domestic-policy advisor and later as senator, Moynihan also had to confront yet another significant set of research findings that pointed in conflicting policy directions. In 1967, planners at the Office of Economic Opportunity, the command post for Johnson's War on Poverty, began developing a series of real-world experiments — the first such in the United States — to test the feasibility of directly supplementing the income of families below the poverty line. Although favored by many experts, this idea — an "income strategy," as opposed to providing a variety of social services to the poor — was expected to encounter political resistance from those who believed such assistance would diminish the incentive for its beneficiaries to support themselves. With no legislative prospect in sight, the Johnson administration decided to try to find out via experimentation whether the income strategy would work or prove its skeptics were right.
Four trials were launched, involving nearly 10,000 families in seven states. Participants received varying amounts of money, and their responses were compared with those of similar families that received nothing. To assess long-term effects, the experiments, which began in 1968, were scheduled to continue through 1982.
Change arrived the very next year, however, when, to everyone's surprise, the new Nixon administration proposed its version of an income strategy, dubbed the Family Assistance Plan. As White House advisor and the program's chief architect, Moynihan had successfully made the case that such an initiative would be not only a more effective way to reduce poverty, but also an improvement over the existing public-assistance programs, which Republicans (and many others) had been criticizing. The so-called "negative income tax" experiments did not play a significant role in the debate within the White House that led to the proposal of the Family Assistance Plan, but they did figure prominently in the legislative debates that followed.
As Congress weighed the Family Assistance Plan, preliminary findings from the first of the Johnson experiments, the New Jersey sample, were released. Initially, they seemed to allay concerns, for employment among families receiving income supplements declined only slightly and much of that occurred among secondary workers (wives and older children), not principal breadwinners. On closer examination, however, it turned out that a large share of the cost of the Family Assistance Plan would go to replace what these secondary workers would no longer be earning. That could be seen as a good outcome, since it meant, for example, that mothers could stay home with young children and older children could spend more time in school, rather than having to contribute to their families' income. But the same finding could also be seen as requiring taxpayers to foot the bill for people who were capable of supporting themselves — not an unambiguous endorsement of an income strategy.
Although Nixon's plan failed to make it through Congress — and Moynihan returned to Harvard — subsequent presidents continued to push for various forms of an income strategy. By the time the Carter administration's version arrived in 1978, Moynihan had become a senator from New York, and the last and largest of the negative income tax experiments — combining data from Seattle and Denver — was beginning to yield results. Although its work-incentive findings were similar to those of the earlier tests, the evidence on the impact of income supplements on family life was new and ambiguous.
One of the problems with existing welfare programs, especially in Moynihan's eyes, was that only single-parent families were generally eligible. Aiding households in which both parents were present would, in theory, eliminate any reason that families might have to boost their incomes by dissolving — or never forming in the first place. But the 1978 Seattle-Denver experiment results revealed that, rather than reducing rates of separation and divorce, income supplements to two-parent families would actually increase rates of dissolution unless the amount was much higher than was likely to be feasible. No one could explain why, though some thought the money served as a kind of publicly financed alimony, enabling women with children to leave their partners. Whatever the reason, an income strategy no longer seemed like a sure way of strengthening low-income families, as Moynihan himself would acknowledge.
POLICY CHOICES AND INCONVENIENT FACTS
Faced with troublesome facts, stubborn and disagreeable opinions, and discordant advice regarding possible solutions, what are well-meaning policymakers to do? In an account of his White House service, Moynihan explained what he thought should have been done with Arthur Jensen's article:
The only responsible course government could take was to proceed as if the hypothesis — it was no more than that nor was it asserted to be — was not true, and to hope to disprove it by a reordering of the environmental influences which, in alternative views, were the essential sources of inequality.
It is important to note that Moynihan was not suggesting that research — even a study as controversial as Jensen's — should be ignored. The findings of a reputable scholar are not to be taken lightly. Indeed, whenever possible, they should be evaluated, exposed to additional analysis — including by competent individuals who might reach different conclusions — and judged candidly and transparently. It might be helpful also to be clear about who finds a given fact "convenient" and who does not. But government is not required to enact policies that align with difficult facts and conclusions, such as those that appeared to flow from the Jensen article a half-century ago.
Policy choices should account for data and studies, but they must also involve interests, beliefs, desires, traditions, calculations of political advantage, religious or ethical views, and, not least, fundamental governing principles. The need for balancing research with interests and values applies across a wide spectrum of issues, economic as well as social, domestic as well as international. Every side of a policy dispute can be expected to adduce innumerable "facts" and research studies to buttress its arguments and advance its preferences. In the end, policymakers must simply strive to ensure that policy choices, if and when they are made, will withstand scrutiny from thoughtful people not aligned with any of the "sides" of the debate, and will survive encounters with the realities they purport to address.
It is also important to remember that constructive policies sometimes emerge from difficult facts and contentious opinions, including measures that may point in the opposite direction. A troublesome piece of research like Jensen's may even provide a welcome impetus for trying to "disprove" its conclusions, as Moynihan suggested. Largely at his behest, after the reaction to Jensen and disappointing early evaluations of Johnson's new Head Start program, the Nixon administration proposed — and Congress agreed to create — a new "National Institute of Education" (today's Institute of Education Sciences) to identify more effective education interventions and programs.
Both the Coleman report and the negative income tax experiments yielded facts and opinions that were uncomfortable for many interested in education and income-support policies. And neither pointed unequivocally to what should be done. Yet they have informed policymaking in these areas for half a century. Indeed, present-day advocates of a "universal basic income" are even mooting new versions of negative income tax experiments.
But not all uncomfortable reports are created equal, even when the scholarship is equally rigorous, and some are simply too problematic to face. Ignoring the facts, however, does not make them go away, nor does it necessarily lead to a better outcome for those on whose putative behalf they're being ignored. As historian James Patterson has written, the unwillingness of many politicians, interest-group leaders, and even scholars to confront the findings of the Moynihan report — despite those findings having been reaffirmed many times since — has set back African-American social and economic progress.
Policymakers — not to mention journalists, scholars, and jurists — would do well to acknowledge the uneasy relationship between facts, opinions, and government actions. But they should also recognize that they have the opportunity, in the face of inconvenient knowledge, to commission new studies, which can lead to new facts, opinions, and government actions. They should also acknowledge that they have the power, and the responsibility, to reverse course when necessary.
In the most fraught political and policy realms, there is no obvious path to right reason. And sometimes "experts" are wrong. "The worst, the most corrupting of lies," wrote Georges Bernanos, a French author Moynihan often quoted, "are problems poorly stated." One might cast the same aspersion on facts that are presented as more definitive than they deserve to be.
When action is called for and choices must be weighed, however, it is still important to consider what is known that might improve the situation. We can acknowledge that facts are frequently shaped by opinions, that opinions resist facts, and that policy choices may be consistent with a wide range of both. But none of that means we should noisily deny facts that we find inconvenient, as some elected officials today are wont to do. Rather, facts should discipline policymaking as much as possible. Many other things will affect decisions — and they should. Knowledge is not enough. But it is better than ignorance.