Scientific misconduct and the responsible conduct of research in science and economics

Abstract

Considered here are matters relating to the responsible conduct of research in economics and science in the United States for the last forty years. In science there was a “late 20th century wave” of scientific misconduct and then a “millennial wave”. For economics in the former era, episodes of honest error and replication failure occurred. Recently plagiarism and data manipulation have been reported. Overall few economists seem to fabricate data, but falsification of data, replication failure, and plagiarism occur. Furthermore, replication failure is the one thing that scientific misconduct and honest error have in common. In economics and compared to the sciences, there have been no misconduct hearings, no economist has been charged with a crime, nor has anyone served time in prison for scientific misconduct. Science and economics seem to be sufficiently self-corrective so that systemic science failure does not utterly thwart scientific progress in the long run.

Notes

¹ For the National Academy study see Fang et al. (2012).

² This author now participates in a research ethics organization within the University of New Hampshire that was created to align its internal grant preparation activities with those required by federal agencies. The UNH program on “The Responsible Conduct of Research and Scholarly Activity” can be found on the university’s website. An informal document compiled by Duke University in January 2014 compares the RCR programs of Columbia, Emory, Harvard, Johns Hopkins, Princeton, Stanford, Vanderbilt, Michigan, Wisconsin-Madison, and UNH. The document is titled: “RCR Reports from Peer Institutions,” compiled by the Duke University Graduate School. See Graduate School (2014).

³ Burt’s unethical activities are considered in Hearnshaw (1979) and Eysenck and Kamin (1981). Kamin (1974) explores matters with the correlation coefficients.

⁴ These are the remarks of the investigating committee as reported in Kohn (1986: 81). The episode can be found in Kohn (1982: 76–83).

⁵ The colleague was Robert Sprague. In response to Sprague’s criticisms, the National Institute of Mental Health was contacted. See Anderson (1988).

⁶ Besides the quote that follows, see Maddox (1987) in Nature and Koshland (1988a, 1988c).

⁷ That research led to headlines in the New York Times like “Misconduct Widespread in Retracted Papers ….” and “Fraud in the Scientific Literature” both from October of 2012. The article from 1 October is written by Zimmer (2012a) and the editorial is from 5 October 2012.

⁸ Johnson (2014). Details of this event are found in Tabuchi (2014).

⁹ The details of this case can be found in three articles by Carolyn Johnson (2011a, 2011b, 2012).

¹⁰ Because Poehlman had received nearly $3 million in grants from the federal government, he was charged as a criminal with fraud. At his sentencing he was given prison time of one year plus one day and two years of probation.

¹¹ Johnson (2002) provides an interesting narrative of this episode.

¹² Fang et al. (2012: 1). There are other studies on misconduct for this millennial period. In one Martinson et al. (2005) surveyed two different groups of scientists. Richard Van Noorden (2011) in Nature published a study on retractions based on Web of Science data from Thomson Reuters. In a theoretical piece which analyzes the strategic decision structure facing the scientist, John Ioannidis (2005) makes the exceptionally bold assertion that most research findings are false.

¹³ For more about Goodwin and Ambrose see Gates (2002) and Mchegan (2002).

¹⁴ For details about Harvard psychiatrist see McDonald (1988). Similarly a dean at Boston University resigned over charges of plagiarism, see Flint (1991).

¹⁵ Naik (2011) provides an overview of the difficulties of replication in biomedical research. Clinical trials are necessary in biomedical science in a way that they may not be in other sciences such as physics and economics.

¹⁶ The reason the estimates could be so different is that theorists recognized two effects of social security that work in opposite directions. An asset substitution effect would be negative and a retirement effect could be positive. So private saving might not fall.

¹⁷ Part of the problem is that one needs to make a complex calculation for the present value of the flow of all present and future social security payments for the country. There can be real differences in how this variable is adjusted for many factors such as remarriage after death of a spouse, the presence of dependent children, and other factors.

¹⁸ The initial story and a follow-up written by Anne Lowery appeared on 16 and 17 April of 2013. There were several responses from Reinhart and Rogoff to Herndon, Ash, and Pollin and many critical responses by Paul Krugman.

¹⁹ See Reinhart and Rogoff (2013a, 2013b). A third response came about a week later, see Reinhart and Rogoff (2013c). See also Lowery (2013a, 2013b).

²⁰ See Cronin (2013b). Charles Giles (2014a, 2014b) of the Financial Times published an overview and a lengthy analysis of data problems in Thomas Piketty’s new book. Then Piketty (2014) posted a reply. There is an interesting report of a miscalculation in projected federal revenues for a 40-year period by Michael Boskin of Stanford in 2003 (Mandel 2003).

²¹ They also appeal to Thomas Kuhn’s conception of normal science and conclude that replication does not fit very well within the reward structure of scientific research.

²² Of 62 authors in the retroactive group, 20 never responded, 22 submitted programs and data, and 20 responded that they could not provide programs and/or data.

²³ Ten years later the replication project was repeated for the Review of the Federal Reserve Bank of St. Louis.

²⁴ A very thoughtful account of the difficulties of replication in economics has been authored by Hamermesh (2007). He describes the replication efforts and incentives in recent economics.

²⁵ Enders and Hoover (2004: 491) note that there have been two public notices of plagiarism in economics in Kylos in 1999 and in the Quarterly Journal of Economics in 1984.

²⁶ Enders and Hoover (2006) repeated the survey in much the same form for a sample of economists who were not editors. The economists were much less cautious than the editors in defining what was plagiarism.

²⁷ Wible and Dietrich (2002) have applied the microeconomics of game theory to clinical trials.