BYINGTON & FELPS: On Resolving the Social Dilemmas that Lead to Non-Credible Science
In our forthcoming article “Solutions to the credibility crisis in Management science” (full text available here), we suggest that “social dilemmas” in the production of Management science put scholars and journal gatekeepers in a difficult position – pitting self-interest against the production of credible scientific claims. We argue that recognizing that the credibility crisis in Management science is at least partly a consequence of social dilemmas – and treating it as such – are foundational steps that can help move the field toward adopting the variety of credibility enhancing practices that scientists have been advocating for decades (e.g. Ceci & Walker, 1983; N. L. Kerr, 1998).
Although we are Management scholars rather than economists, we suspect that the social dilemma dynamics we point out (and the solutions we propose) are relevant for improving the credibility of claims produced by many fields (e.g., economics, sociology, anthropology, psychology, criminology, education, political science, medicine, etc.). As such, we are grateful for the invitation from The Replication Network to share a summary of our article for your consideration.
Credibility Problems in Management Science
The claims of primary studies in Management cannot be fully relied upon, as evidenced by the fact that a) results fail to replicate much more often than they should, and b) attempts to verify and replicate prior claims rarely appear in the literature (Hubbard & Vetter, 1996).
There is reason to believe that the weak replicability of Management findings may be the result of four sets troublingly prevalent researcher behaviors (see full manuscript for evidence of prevalence):
— Unacknowledged “Hypothesizing After the Results are Known” (N. L. Kerr, 1998);
— Data manipulation (also known as p-hacking), which involves exploiting researchers “degrees of freedom” – e.g. adding / dropping control variables, dropping uncooperative data points / conditions, using alterative measures / transformations – to find desired results (Goldfarb & King, 2016);
— Data fraud, which involves changing data points or generating data wholesale (John, Loewenstein, & Prelec, 2012);
— Data hoarding, which involves an unwillingness to share data or research materials that would allow others to verify whether one’s data is consistent with one’s published conclusions (Wicherts, Bakker, & Molenaar, 2011).
As demonstrated in studies such as that of Simmons, Nelson, and Simonsohn (2011), such practices can dramatically increase the likelihood of producing “statistically significant” (but ultimately erroneous) findings.
Drivers of Non-Credible Research Practices
We argue that the reason these undesirable research behaviors are so prevalent in Management is that engaging in such behaviors can be beneficial for one’s career, since such behaviors facilitate the production of highly citable (i.e. novel, theory adding, statistically significant) research claims likely to be publishable in high status journals. Of course, engaging in these research behaviors comes with some risk of detection, but the current lack of verification / replication efforts would seem to make the chance of detection low. Thus, scholars are in a social dilemma, where what is good for them individually (i.e. producing highly citable claims) is at odds with what is good for society/science as a whole (producing credible, replicable claims).
There are a variety of journal practices that would significantly decrease the career benefits associated with non-credible research practices, and thus lead to more credible Management science. They include:
— Frequent publication of high-quality strict replications via dedicated journal space, distinct reviewing criteria for replication studies, provision of replication protocols, and crowd-sourcing replication efforts;
— Enabling robustness checks through in-house analysis checks and altered data submission policies;
— Enabling the publication of null results through registered reports and results-blind review;
— Adopting Open Practice article badges (Center for Open Science, 2015).
However, adoption of these practices has been slow. We propose that one possible reason is that a journal that “sticks its neck out” and adopts these credibility supportive practices is likely to see its status decline. For example, null findings and replications are rarely cited (Hubbard, 2015), and thus publishing them can reduce a journal’s impact factor. Similarly, requiring scholars to submit their data when competitor journals do not have such a requirement will make the “purist journal” a less attractive publication outlet for scholars, potentially reducing their pool of highly citable submissions. Indeed, each of these credibility enhancing journal practices are likely to lead to research that is both more reliable and less citable. This means that journal gatekeepers (editors and reviewers) are themselves trapped in a social dilemma, where what is good for the journal’s status (i.e. high impact factor relative to “competitor journals”) is at odds with what is good for society/science as a whole (i.e. adopting credibility enhancing practices that help ensure more reliable claims).
Resolving the Social Dilemmas
Fortunately, social science has accumulated great deal of knowledge about how to resolve social dilemmas (Kollock, 1998; Van Lange, Balliet, Parks, & Vugt, 2013). Specifically, we suggest three structural social dilemma solutions, and two motivational social dilemma solutions.
Structural social dilemma solutions involve changing the incentives for journal gatekeepers (Messick & Brewer, 1983). We suggest the following structural social dilemma interventions:
— Define small peer journal groups: A prerequisite for conditional pledges (below) and other social dilemma solutions is identifying a population of peer (i.e. “competitor”) journals.
— Conditional pledges by editors: These are public pledges to adopt certain credibility supportive journal practices if a substantial portion of peer journals also agree to the pledge. This approach is meant to mitigate “relative status costs” of a journal adopting credibility supportive practices.
— Reviewer pledges: Credibility-minded reviewers could themselves publically pledge to (only) review for journals that adopt credibility supportive journal practices to create an incentive for editors to sign onto a conditional pledge with their peer journals.
Motivational social dilemma solutions increase the desire to generously cooperate with others without changing the underlying incentives (Messick & Brewer, 1983). We suggest the following motivational social dilemma interventions:
— Increase multi-journal communication: Editors are more likely to cooperate with other journals in adopting credibility supportive journal practices if they discuss the field-level benefits of doing so face-to-face with their peers (i.e., other editors).
— Inject a moral frame: Journal editors are more likely to adopt credibility supportive journal practices when such practices are framed as a moral imperative.
Across many fields, there is a growing appetite for improving the way science is done. The social dilemma solutions presented in the article build on the belief that the best hope for resolving the credibility crisis in science is in pragmatic (re)consideration of scholars’ and journal gatekeepers’ incentives for producing credible scientific claims. Until then, we are merely rewarding A while hoping for B (S. Kerr, 1975).
Byington, E. K., & Felps, W. (forthcoming). Solutions to the credibility crisis in management science. Academy of Management Learning & Education.
Ceci, S. J., & Walker, E. (1983). Private archives and public needs. American Psychologist, 38(4), 414–423. http://doi.org/10.1037/0003-066X.38.4.414
Center for Open Science. (2015, January 24). Badges to acknowledge open practices. Retrieved January 26, 2015, from https://osf.io/tvyxz/wiki/1.%20View%20the%20Badges/
Goldfarb, B. D., & King, A. A. (2016). Scientific apophenia in strategic management research. Strategic Management Journal, 37(1), 167–176.
Hubbard, R. (2015). Corrupt research: The case for reconceptualizing empirical management and social science. Newcastle upon Tyne, UK: Sage.
Hubbard, R., & Vetter, D. E. (1996). An empirical comparison of published replication research in accounting, economics, finance, management, and marketing. Journal of Business Research, 35(2), 153–164. http://doi.org/10.1016/0148-2963(95)00084-4
John, L. K., Loewenstein, G., & Prelec, D. (2012). Measuring the prevalence of questionable research practices with incentives for truth telling. Psychological Science, 23(5), 524–532. http://doi.org/10.1177/0956797611430953
Kepes, S., Banks, G. C., McDaniel, M., & Whetzel, D. L. (2012). Publication bias in the organizational sciences. Organizational Research Methods, 15(4), 624–662. http://doi.org/10.1177/1094428112452760
Kerr, N. L. (1998). HARKing: Hypothesizing after the results are known. Personality and Social Psychology Review, 2(3), 196–217.
Kerr, S. (1975). On the folly of rewarding a, while hoping for b. Academy of Management Journal, 18(4), 769–783. http://doi.org/10.2307/255378
Kollock, P. (1998). Social dilemmas: The anatomy of cooperation. Annual Review of Sociology, 183–214.
Messick, D. M., & Brewer, M. B. (1983). Solving social dilemmas: A review. Review of Personality and Social Psychology, 4(1), 11–44.
Simmons, J. P., Nelson, L. D., & Simonsohn, U. (2011). False-positive psychology: Undisclosed flexibility in data collection and analysis allows presenting anything as significant. Psychological Science, 22(11), 1359–1366.
Van Lange, P. A. M., Balliet, D. P., Parks, C. D., & Vugt, M. van. (2013). Social dilemmas: Understanding human cooperation. Oxford, UK: Oxford University Press.
Wicherts, J. M., Bakker, M., & Molenaar, D. (2011). Willingness to share research data is related to the strength of the evidence and the quality of reporting of statistical results. PLOS ONE, 6(11), e26828. http://doi.org/10.1371/journal.pone.0026828