Should You Publish?

A Question in Scientific Ethics

You are a researcher with a theory and a clever way of testing it. The test supports the idea. You believe that publishing your results will have undesirable real world effects. Should you publish?

Minimum Wage

For most of my life most economists believed that minimum wage laws decreased employment of unskilled workers, some that the laws should be abolished, a view that was not merely rejected by popular wisdom but, for everyone but economists, outside the range of respectable opinion.

A story from Leo Rosten, a friend of my parents¹ and a prominent author:

On one occasion, he asked an MIT economist about minimum wage laws and got the conventional answer, that they reduce employment opportunities for unskilled workers by pushing up the cost of hiring them. He asked if that was the view of most economists and was told that it was. He asked why, in that case, economists did not make an effort to inform the public of the problem.

“I guess we're afraid of sounding as if we agree with Milton Friedman.”

David Card and Alan Krueger thought of a possible exception to that conclusion, a reason why a small increase in the minimum wage might under some circumstances produce no reduction in such employment, might even produce an increase. They tested that conjecture by looking at a situation where one state had raised its minimum wage and an adjacent state had not, and found evidence of an increase rather than a decrease in employment.² They reported theory and evidence in an article that got a great deal of attention.

Suppose the authors saw their conclusion, as I do, as an exception to a general rule that in most cases held. Further suppose they believed that a large increase in the minimum wage would have unambiguously bad effects.³ Finally assume that they believed that the publication of their article would, by making a position that almost everyone else already believed in respectable among economists, make large increases more likely, as it did.

Should they publish?

IQ

A researcher believes that IQ is in large part heritable and has a substantial effect on life outcomes, that higher IQ correlates with higher wages, less unemployment, more stable marriage.⁴ The logic of Darwinian evolution implies that populations that have evolved in different environments will have different distributions of heritable characteristics, a conclusion supported in the case of easily observed characteristics such as height or skin color. He concludes that races as conventionally defined may well differ in the distribution of IQ, looks for evidence and finds it.

Further assume that he believes publishing these results will have a bad effect. People who are racially prejudiced will jump from “the distribution of IQ is different in different races” to “all blacks are stupid” and act on that belief, making the world a worse place. Because differences in heritable characteristics provide a possible explanation for differences in outcomes by race, people will use his result as an excuse to ignore differences due to discrimination, past injustice, or some other cause, reducing the political pressure to do something about those problems.

Should he publish?

Climate

A researcher finds evidence that climate change will be slower than currently believed or that its results will be less bad. Assume he still believes that climate change is a serious problem and we should be doing more than we are doing to prevent it. Publishing the result of his research will reduce the political pressure to do so.

Should he publish?

Covid

Someone notices a mistake in the theoretical analysis of the pandemic, a mistake that results in an overestimate of how many people would have to have had the disease to get to herd immunity. Pointing out the mistake would make Covid appear less threatening and so reduce efforts, individual and governmental, to fight it, which he believes are already too low. Should he publish?

This, like my first example, is real.⁵ The following, about the paper, is from the editors' blog of Science magazine:

we were concerned that forces that want to downplay the severity of the pandemic as well as the need for social distancing would seize on the results to suggest that the situation was less urgent. We decided that the benefit of providing the model to the scientific community was worthwhile.

Their view, pretty clearly, was that hiding true information might be justified in some such situations but in this case wasn’t.

My Conclusion

The strongest case against publishing is for the minimum wage article, not because there was anything wrong with the research but because, since the conclusion was one that many people already wanted to believe, a single piece of evidence in favor would outweigh much evidence against.⁶ That was a reason for the authors to be careful to qualify their conclusion, to point out reasons it might be wrong. It was not a reason not to publish either their evidence or their argument. The same would have been true for the case of IQ a century ago but not for that or the other two today. In each of those cases, failing to publish would mean concealing evidence against the currently popular view, evidence more likely to be minimized than exaggerated.

The argument for publishing, in all four cases, is similar to the argument for freedom of speech and the press. A world where someone has the power to suppress true information he disapproves of, whether because the information is in his head or because he is a government censor, is less likely to produce true conclusions. Better to let the arguments be fought out in the marketplace of ideas.

I don’t want to overstate my conclusion. If I come up with a clever way that a bright high school student could engineer a plague virus or an atomic bomb in his basement I won’t publish. But for ordinary controversies, questions with reasonable people on both sides, suppressing true information makes it less likely that truth will emerge, especially if the true information is evidence against the current popular view — which is the kind of information most likely to be suppressed. And it is all too easy for someone on one side of a controversy to conclude that there are no reasonable people on the other side.

My Biases

Before ending I should concede my biases. The Card-Krueger argument depends on the employers of unskilled labor being monopsonists, each the only employer of such labor in his area. That strikes me as unlikely in the US at present, the empirical evidence they offered real but weak. Since both the theoretical point and the evidence were true the authors were right to publish, but doing so probably had bad consequences.⁷ On the other hand my view of climate change, as regular readers are likely to realize, is that the current consensus may well be mistaken, exists in part because many people have incentives to bias the results they publish to support it, minimize or conceal results that undercut it.

1. There is a sketch of my father in his People I Have Loved, Known or Admired.”

2. For a description and critique of the empirical part of the article, see “No, Krueger Didn’t ‘Prov[e] that Raising the Minimum Wage Doesn't Increase Unemployment’”. See also “The Minimum Wage and Monopsony” by David Henderson. The theoretical point, which depends on the employers of minimum wage workers being monopsonists on the market for unskilled labor, was made much earlier by George Stigler (“The Economics of Minimum Wage Legislation” AER Vol. 36, No. 3. (Jun., 1946), pp. 358-365)

3. “It doesn't mean that if we raised the minimum wage to $20 an hour we wouldn't have massive problems, if we enforced it.” An Interview With David Card.

4. Roughly the results reported in The Bell Curve.

5. I explained the argument in an old blog post. A preprint of an article making the same point in much more detail.

6. To check that, Google “David Card minimum wage” and compare the number of hits that critique the article to the number that endorse it.

7. A view that Paul Krugman shared in 1998, five years after the article was published: “What is remarkable, however, is how this rather iffy result has been seized upon by some liberals as a rationale for making large minimum wage increases a core component of the liberal agenda…” but rejected in 2015. (Paul Krugman on the Minimum Wage: 1998 Vs. 2015)