There are three motivations to replicate someone else’s work. First, you learn statistics and methods. Second, you provide a service to the community by advancing knowledge and validating results. Third, and most importantly for many, if you add value, you can get it published. Here are some success stories from the social sciences about papers that got published by replicating existing work.
More and more often, stories appear about retraction of papers due to misconduct, for example in the journal of Infection and Immunity (see New York Times), when several papers had to be retracted because the author had made up results. Another paper in the British Journal of Social Psychology was retracted due to fabrication of data (see Retraction Watch). The Conflict Resolution Quarterly journal retracted an article in which the author had combined two other articles into one, claiming it was his own work (see Retraction Watch). A study on childbirth at home versus in the hospital in Feminism & Psychology was retracted after conclusions were not actually supported by the results – women were discouraged from home birth based on inadequate evidence (see Retraction Watch). After the twitter hype about misconduct in the academic world (#overlyhonestmethods) the discussion on replication, reproducibility and data sharing is attracting even more attention.
Journal expect novel research
However, we will only ever find out about misconduct if we reproduce more work. Often, motivation for this task is not always high. Journals expect novel research, and especially graduate students get pushed by their institutions to produce and publish original work as soon as possible. So why would anyone take the time to replicate?
The way social sciences work today, the main motivation can only be publication. Replicating work should not only lead to retractions, but journals should invite replications – with added value – to be published as a stand-alone paper. I’m trying to push my students in the Cambridge Replication Workshop to get to a point when they add robustness checks, new variables and interactions, updated data, multiple imputation and improved methods to their replication project so they can write a publishable project (see Gary King‘s tips on how to write a publishable article based on replication).
Replication articles: political science
This is a selection of interesting replication articles that made it into journals in political science in the last years.
- Understanding the Effect of Political Advertising on Voter Turnout: A Response to Krasno and Green, by Michael M. Franz, Paul Freedman, Ken Goldstein and Travis N. Ridout, The Journal of Politics, Vol. 70, No. 1, Jan., 2008.
- Blood and Soil? Resource Scarcity and Internal Armed Conflict Revisited, by Ole Magnus Theisen, Journal of Peace Research, Vol. 45, No. 6, Nov., 2008.
- The Economic Effects of Constitutions: Replicating—And Extending—Persson and Tabellini, by Lorenz Blume, Jens Müller, Stefan Voigt and Carsten Wolf, Public Choice, Vol. 139, No. 1/2, Apr., 2009.
- Party Identification in Emotional and Political Context: A Replication, by Francis Neely, Political Psychology, Vol. 28, No. 6, Dec., 2007.
- Self-Prophecy Effects and Voter Turnout: An Experimental Replication, by Jennifer K. Smith, Alan S. Gerber and Anton Orlich, Political Psychology, Vol. 24, No. 3, Sep., 2003.
- The Two Presidencies, 1984-98: A Replication and Extension, by Scot Schraufnagel and Stephen M. Shellman, Presidential Studies Quarterly, Vol. 31, No. 4, Dec., 2001.
If you have more articles I could add to this list, or a list of wider social science replication papers, please comment below.