Let me start by asserting that I am a big fan of science. Contemporary society has only been possible through a radical cognitive revolution that encompassed inductive reasoning, rational thought, and core scientific methods. Such a conceptual revolution typically is associated with Francis Bacon (b. 1561), known as ‘the father of empiricism,’ though earlier but (alas) temporary conceptual breakthroughs along similar lines occurred during the Islamic Golden Age almost a millenia earlier. Obviously, many others contributed to this ‘revolution.’

Nor should we forget the long  struggles to free society from its preferences for understanding the world about us premised on superstition or revealed truths. Think about Galileo’s (b. 1564) struggle with Pope Urban VIII over his heliocentric theory of our solar system and many other conflicts all the way down to the 1920’s Scopes trial that convicted a Tennessee teacher of exposing his students to the dangerous theory of evolution. Even today, entrenched right-wing ideologies ridicule and attack those products of our best scientific processes when research challenges sacred preconceptions and values. Consider HHS Secretary Robert F. Kennedy’s dismissal of accepted  medical research in favor of agenda- driven beliefs as he directs our nation’s premier health agency. In case you missed it, vaccines generally do work. Good science confirms that.

Moreover, I have spent much of my professional life in the academy. Admittedly, I was never a scholar by disposition but, as the saying goes, most of my friends and colleagues were of that ilk. Being the Associate Director of a top university- based multi- didciplinary research entity for many years afforded me great opportunities to see rigorous analytical methods being applied to our more pressing social policy challenges. I remain grateful for being able to associate with the best and the brightest.

Nevertheless, even these best and brightest can succumb to errors derived from self-interest and cultural myopia. That is only human. Let me only touch on two, shall we say, methodological conundrums.

Economic rationality. I spent the vast majority of my professional life surrounded by economists. In general, I liked them even if they were a testy and disputational lot. Surprisingly, they seemed to like me, significantly more than did my social work colleagues. Nevertheless, I am moved to quibble with a dominant pillar of economic research.

Somewhere in the 1960s, there was a significant revolution in what would be considered good science within the discipline of economics. Descriptive work suddenly was replaced by quantitative techniques and a highly theoretical approach to the discipline. Higher mathematics and abstract manipulations replaced broader or more balanced attempts to understand our economic world or society in general.

Fancy econometrics became the sine qua non for acceptable analysis and, most critical to academics, publication in peer reviewed journals. Practitioners of the dismal science wanted their discipline to look more like physics. Complex equations were the way to do that, and to further separate economic members of the academy from those in the softer social sciences and, of course, the public. Obscurity was its own reward.

These fancy econometric explorations have their uses, don’t get me wrong. But they also have built-in limits, especially when one seeks to apply results to the real world. Over time, it became clearer to us doing applied work in the academy that newly minted Ph.D.s in economics were interested in theory, not practice. Even the character of the doctoral students had visibly changed, increasingly being dominated by Asian and Eastern European countries which produced math and computer whizzes who knew little about institutional realities, key historical lessons, or human idiosyncrasies. A close economics colleague at the time bemoaned this trend that we all observed.

To my mind, the core issue with the new economics boiled down to this. Increasingly, at the start of each Institute (the Institute for Research on Poverty) brown bag, the presenter would go through a litany of assumptions that the audience must accept in order to make the math work. Essentially, these assumptions turned humans into stick figures only motivated by achieving one’s optimal utility or wellbeing as measured in dollars. All cultural factors, institutional idiosyncrasies, and emotional intrusions were dismissed. People were reduced to mini calculators operating in perfect markets that dismissed annoying issues like externalities or asymmetrical information flows. In such an ideal world, with everything monetized into dollars, the math worked very well indeed. Economics did look like physics until you looked closely.

I considered this quirk early in my career. This tendency to reduce humans to homo- economicus struck me as overly stylized and simplistic. I decided to test my concern. There was a poverty policy tactic generally termed wage-bill subsidies. Essentially, the government would subsidize the wages of harder to employ job seekers, thus improving their labor market prospects, at least according to economic theory.

After all, classic economic theory had reduced decionmakers to utility- maximizing automotons. Hiring supervisors would see a reduced  wage bill (the outlay to secure a unit of labor services) and favor the subsidized individual. That was obvious … no? The problem arose when it became clear that such programs were significantly under- subscribed. In short, the subsidies went under- utilized. Most economists were flummoxed.

So, I conducted an unorthodox study of this reform strategy by doing something unconventional. I and my team did a series of paper surveys, personal interviews, and focus groups. That is, we talked to decision-makers … real people! That was something respectable economists seldom did. Good thing I was not one of those.

Amazingly, our results suggested that people simply are not human calculators nor utility- maximizing automatons. Their reasoning is far more complex than economists assume, too complex to cover here. One perverse outcome of my research suggested that wage subsidies could hurt job seekers in some circumstances … a counter- intuitive outcome. Some hiring supervisors saw the public subsidy as a red flag, suggesting that the cost of taking such a person might well exceed the value of a subsidy. In general, however, the hiring process involves more than estimating future utility, a fairly subjective endeavor indeed. Feelings, intuitions, and preconceptions enter into such decisions.

The bottom line, for me, was this. Humans are more than calculating stick- figures that go around calculating monetized utility all day. They are, inconveniantally, quite human.

The ‘gold standard’ of social science research. Among my other observations of methodological conundruns stood a counter- intuitive blasphemy. Good science might just confound good policy making, at least in some situations. Now, what could that possibly mean?

This story goes something like the following: I spent a year on leave from the University of Wisconsin working at the Department of Health and Human Services (HHS). This was during the early years of the Clinton Administration. My primary reason for this segue into the real world was to work on welfare reform … a hot issue at the time. Now, remember this, welfare reform had been described by a previous Secretary of HHS ((Joseph Califano) as the Mideast of Domestic Policy. These issues were so contentious that I oft remarked that I knew I was getting closer to the truth when no one agreed with me. It was not an avocation for the faint of heart but I thought it grand fun.

Planning for national welfare reform was to be headed by two Harvard Profs (David Ellwood and Mary Jo Banes) who had accepted high political positions at HHS. In addition, a number of academics and research types had been recruited to spend time in D.C., as I had been. The point of all this was to guarantee that Clinton’s welfare reform would be guided by the best research possible, not by seat-of-the-pants ideology. Science, not politics, would prevail.

I recall the phone call from Ann Segal asking how quickly I could get to Washington. I told her I would have to finish my semester teaching responsibilities and take care of my mother’s affairs (she had passed that Spring of 1993). Ann suggested I get to D.C. as promptly as I could … they would be moving quickly on the reform agenda. Surely I wouldn’t want to miss working on this policy vessel before it was launched. It turned out I had nothing to worry about; I saw on day one that the reform agenda was yet stuck in dry dock and would remain there for some time.

Unlike Trump’s menagerie, Clinton’s team was filled with really smart people. These truley were the best and brightest who expected, more like demanded, decisions to be based on excellent science. By ‘excellent,’ they wanted support based on rigorous observational studies or, preferably, by interventions supported by gold standard experimental studies. A ‘gold standard’ study employed  randomly assigned subjects into one or more experimental groups along with a control group. This approach would ensure that extraneous factors would not confound any observed effects. Other protocols are essential but random assignment is key.

Now, here’s the rub. Random assignment studies worked best for distinct, segmented interventions where individual subjects could be assigned to separate treatments and a control group. At the time, we had plenty of those studies to examine. The difficulty, in my eyes, was that the true impacts (as opposed to spurious results from non-randomized studies) usually were quite small even though statistically significant. You could reduce welfare use and raise family income but not by much.

While scientists can get excited by statistically significant results, policy makers demand something more … substantive significance. Take the following situation as an example! Clinton said reform must be budget neutral. He also implied that his package would include a time- limit on receiving assistance. Thus, we were forced to come up with interventions that would land large portions of the welfare population in jobs, and decent jobs at that. If not, public jobs would be required and that would be more expensive than just giving recipients a monthly check. There would go budget neutrality. Nor was there any political will to kick people off assistance when they hit their time limit if they had played by the rules.

But the best science we had said that existing reform initiatives (those that had been rigorously tested) would have marginal impacts, outcomes well short of what was required to be cost neutral while keeping to a time- limited program structure. I quickly realized that the best research was leading us astray. How was that?

The so-called best research focused on topics that lent themselves to gold standard randomized experiments. That is, they typically focused on one or two isolated interventions. What I (and a few others) realized was that we needed radical changes that would transform the entire culture of welfare offices as well as radical changes in basic programmatic concepts. But there was no existing research support for such radical ideas, especially in a policy area rife with political turmoil. Washington was a difficult town in which to effect dramatic change. More to the point, changing the culture of larger systems presented so many logistical and methodological challenges that they were seldom, if ever, subject to rigorous study.

At one point, I and another academic colleague on leave (Rebecca Maynard from the University of Pennsylvania) argued that Clinton should not propose a national package. Rather, make the states a true laboratory of reform with support for dramatic transformation in the way business is done. Perhaps in that environment, some states could experiment dramatically while others would follow the best examples supported by existing research. In effect, states would become the unit of analysis in a national experiment.

But that was not to be. Clinton had promised to end welfare as we knew it. The planning process plodded on. By the time the President’s plan was released, it was too late. The Republicans under revolutionary House Republican leader Newt Gingrich would soon take control over the reform agenda. And they had no interest in research.

Concluding thoughts. I have been out of the policy and research games for some time now. But I was moved to comment on a couple of my old concerns after reading The Road to Freedom by Nobel Laureate Joseph Stiglitz. He took apart contemporary classical economics while attacking the use of utility- optimizing stick men at the center of so- called rigorous analytics. I could not agree with him more.

One final thought on my assertion that the best research can thwart good policy. Sometimes after my time in D.C. I was asked to participate on a National Academy Expert Panel to look at the best ways to evaluate the national welfare reform passed in 1996. I had an opportunity to express my concerns and found the panel sympathetic. The final report reflected that but welfare would soon disappear as a major issue.

The thing is that, despite all limitations, science and inductive rationality are the most promising avenues toward progress. But we must always remember that science is not ultimate truth. We must continue to assess our techniques and question our assumptions. The search for truth never ends.

Perhaps the AI machines that will soon replace humans in doing high level cognitive tasks will do a better job.