Interdisciplinarity, Heritability, and Public Policy
We are living in times of hyperspecialization. Since the Industrial Revolution (1750–1850) we have witnessed an explosion of technological inventions and an enormous accumulation as well as dissemination of knowledge. As a result, it is no longer possible for a single human being to keep up with the latest trends in widely diverse scientific fields such as geology, mathematics, psychology, genetics, and neuroscience. Let alone that it would be possible to master proficiency in one or more of these fields while at the same time being one of the greatest painters of all time—Leonardo da Vinci epitomized this Renaissance humanist ideal, and with him it died.
Nowadays, it is even increasingly difficult (if not impossible) to keep up with the recent trends and discoveries within a specific scientific field such as mathematics, psychology, sociology, or neuroscience. To give just one example: There are today 305 economics journals listed in the Web of Science directory. Those journals include generalist journals such as the American Economic Review and the Quarterly Journal of Economics, but also highly specialized journals from various subfields within the domain of economics (such as accounting, finance, marketing, macroeconomics, etc.). Some of those journals only publish a few articles a year (the Journal of Labor Economics, for example, publishes about 20 articles a year), others publish articles at a dazzling speed (for example, the flagship of the economics profession, the American Economic Review, publishes more than 200 articles a year). Taken together, there are nowadays about 20,000 scientific articles being published about economics each and every year! If one would also consider books, articles that appear in scientific journals which are not indexed in Web of Science, or (un-reviewed) working papers being disseminated through personal websites and servers such as SSRN (the Social Science Research Network), the total scientific output is even greater. Even if one only considers the "top" journals in the field (in economics that would be the American Economic Review, Econometrica, the Quarterly Journal of Economics, the Journal of Political Economy, and the Review of Economic Studies), we are talking about up to 400 articles a year. Little time is left for anything else if one wishes to carefully read each and every of those articles.
So, nowadays, most scientists are no longer economists, sociologist, psychologist, neuroscientists, or physicists (except maybe on their official letterheads), rather they are accounting researchers, family sociologists, cognitive psychologists, clinical neuroscientists, or astrophysicists. At least as long as they are interacting with somebody from outside their discipline; on a day to day basis they are probably working in even smaller niches and likely to describe themselves as financial accounting researchers, theoretical astrophysicists, and so on and so forth. (Just to give you an idea: Web of Scienceindexes 18 accounting journals, adding up to more than 500 articles being published every year.)
Hence, being truly engaged in more than one scientific field requires considerable effort and devotion, especially if there traditionally has been little overlap between these fields (gaining profound knowledge in economics and, for example, biology is arguably harder than combining mastery of economics with expertise in mathematics). It is thus no wonder that there nowadays is an increase in multidisciplinary scientific teams while at the same time there is a decrease of truly multidisciplinary individual scientists. Of course, there are still (some) scientists who transgress scientific boundaries and who truly manage to impact diverse scientific fields, but this hardly hides the fact that interdisciplinarity is nowadays unattainable for (most) individual scientists, even as a mere ideal.
So clearly, we must applaud any attempt being undertaken to bridge the gap between different scientific disciplines. Being myself an economist/social scientist with a genuine interest in neuroscience and genetics, I look with both interest and skepticism at articles that try to enhance our body of knowledge by combining theories and empirical findings from different scientific fields. In a number of cases (for example, the combining of economics with psychology) such attempts have resulted in interesting new discoveries and (theoretical) insights (powerfully illustrated in this case by the fact that psychologist Daniel Kahneman was awarded the 2002 Nobel Prize in Economics). Sometimes, such enterprises produce new empirical evidence and are likely to turn out to be worthwhile scientific endeavors in the future; such as is the case for the recently emerged field of neuroeconomics (combining insights from economics and neuroscience). In other cases, the value that knowledge form one scientific field can add to another has yet to be proven (for example, despite huge claims to the contrary, it has yet to be shown that genetics will really be able to add anything meaningful to our understanding of social, political, and economic processes).
Unfortunately, scientists who wander off their immediate field of expertise and explore lesser-known territory, occasionally also encounter the boundaries of their knowledge, fall prey to misinterpretations, and draw unwarranted or even erroneous conclusions. It needs little explanation that in such cases scientists do more harm than good. The dissemination of misinterpreted results, wrongful deductions, or erroneous conclusions can hamper scientific progress, or even be dangerous.
This should be a minor (if not completely nonexistent) problem as long as the peer review system is working properly. Ever since the first publications of scientific articles in the Philosophical Transactions of the Royal Society in 1665, scholarly and scientific publishing has relied on peer-review to ensure the quality of the published work. That is, a paper that is submitted to a scientific journal such as Nature, Science, the American Economic Review, or any other of the many thousands that are being published nowadays, will get scrutinized by other scientists who are experts in the field before the editor of the journal will consider it for publication. Although practices vary across scientific fields, it is nowadays extremely rare that a paper gets published without the author(s) making considerable adjustments to it following the comments from the (most of the time two or three) reviewers that were selected to referee the paper (either by the author(s) or, more commonly, by the editor). Thus, with few exceptions, all scientific papers are quality-controlled before they are published by at least one scientist who is considered an expert in the field. To the extent that the review process itself would be errorless, one could thus expect scientific publications to be "correct" (at least as to our current understandings). Unfortunately, however, it has been noted by many that the review process itself is flawed and spurred with many problems. Most importantly for the discussion at hand, there exists substantial empirical evidence that suggests that plenty of good papers get rejected by reviewers and, even more importantly, that plenty of bad papers get published (in "low-quality" journals, but just as well in recognized "top" journals). Although blatant mistakes, omissions, or errors are, in all likelihood, very exceptional, more subtle flaws are presumably quite widespread.
It is easy to imagine that the likelihood that a peer-reviewed scientific publication contains mistakes (or is overly simplistic or to foregoing in interpreting findings from previous research) is greater when that publication deals with a topic that lies (somewhat) outside of the (traditional) scope of the journal in which it is being published. For example, The Journal of Law and Economics probably would be (very) unlikely to contain mistakes on legal issues, slightly more likely to have mistakes on sociological theory, and significantly more likely to contain mistakes when it comes down to research on neuroscience or genetics. There can be no doubt about the fact that the editors from journals such as the American Economic Review and the Annual Review of Sociology (or other such journals) are much less knowledgeable about, for example, neuroscience and genetics than about economics and sociology; they will also be much less knowledgeable about these topics than the editors from journals such as PLoS Genetics or Nature Genetics. Furthermore, also the selected reviewers will presumably be less knowledgeable (and thus less likely to detect mistakes) when a paper surpasses scientific boundaries and/or is submitted to a journal for which the publication of such papers is unusual. After all, where would the editor of, for example, the American Economic Review need to find suitable reviewers for an article that combines economic theory with genetics? I do not want to suggest that proper refereeing is impossible in such cases, but it is harder, and mistakes are more likely to go undetected.
A recent article published in Kyklos (a journal that publishes article on economic and social issues), unfortunately, nicely illustrates my case. In an article entitled, "Biology, Immigration, and Public Policy", Gregory B. Christainsen (a professor of Economics at California State University) draws from recent research in biology (mainly genetics) to point out that 'mass migration does have […] very large risks' (p. 175). Many things could be said about this article and the conclusion it reaches at. I like to focus on one particular issue; the fact that Christainsen seems to suggest that heritability estimates provide relevant information for policy makers.
Christainsen first reiterates the basic findings from behavioral genetics: (a) all human behavioral traits (such as IQ) are heritable, (b) the effects from the "shared environment" (e.g., family) on such traits are smaller than the effects of genes, and (c) a substantial portion of the variation in complex human behavioral traits is not accounted for by the effects of genes or families (see Turkheimer, 2000)—so far so good. However, Christainsen subsequently errs by taking these results to imply that genetic influences are more important and more irreversible than environmental ones. More importantly, his discussion seems to imply that social policy can do little to ameliorate inequality of achievement if traits (such as IQ) are largely heritable.
Christainsen is of course not the first—and, unfortunately, he is probably also not going to be the last—to connect heritability measures (for example of IQ) with social policy. Ironically, from a policy perspective, heritability measures are completely irrelevant. After all, variance decompositions do not yield estimands of policy relevance. (Heritability estimates decompose the variance in a particular trait into genetic and environmental factors; that is, they tell you how much of the variance in a particular trait is being explained by genetic and environmental factors.) It is bitter irony that economists have been telling other scientists for years that the policy-relevant effect of a variable is properly measured by its norm of reaction (i.e., its regression slope), not by how much variance the variable contributes (i.e., its contribution to R2). Already in 1979, this was spelled out in a dazzling clear way by the economist Arthur S. Goldberger. In a marvelous article, published in Economica, Goldberger wittingly demonstrated the absurdity of deriving policy implications from heritability estimates.
In 1977, a study of more than two thousand pairs of twins indicated that genetic factors play a huge role in determining an individual’s earnings capacity, which led the renowned psychology professor Hans Eysenck to proclaim that it 'really tells the [Royal] Commission [on the Distribution of Income and Wealth] that they might as well pack up.' Such a conclusion is of course completely unwarranted (conflating high heritability estimates with irreversibility of genetic influences; that is, genetic determinism). As wittingly noted by Goldberger (1979, p. 337), 'A powerful intellect was at work. In the same vein, if it were shown that a large proportion of the variance in eyesight were due to genetic causes, then the Royal Commission on the Distribution of Eyeglasses might as well pack up. And if it were shown that most of the variation in rainfall is due to natural causes, then the Royal Commission on the Distribution of Umbrellas could pack up too.'
It is unfortunate that as recently as last year, Charles F. Manski (another economist who does understand genetics) brought the article by Goldberger back to the attention of the economics community. In an article that was published in the Journal of Economic Perspectives, Manski reiterated many of the arguments originally put forward by Goldberger to demonstrate that high heritability estimates are uninformative to assess the potential effectiveness of social interventions. Both articles are in fact "must reads"; not just because they are informative and insightful, but also because they are well written and fun to read (which are both exceptional qualities of scientific writings in general, and economics literature in particular).
- Christiainsen, G. B. (2012) Biology, Immigration, and Public Policy. Kyklos, 65: 164–178.
- Goldberger, A. S. (1979) Heritability. Economica, 46: 327–347.
- Manski, C. F. (2011) Genes, Eyeglasses, and Social Policy. Journal of Economic Perspectives, 25: 83–94.
- Turkheimer, E. (2000) Three Laws of Behavior Genetics and What They Mean. Current Directions in Psychological Science, 9: 160–164.