The reliability of replications: a study in computational reproductions
Abstract
This study investigates researcher variability in computational reproduction, an activity for which itis least expected. Eighty-five independent teams attempted numerical replication of results from anoriginal study of policy preferences and immigration. Reproduction teams were randomly groupedinto a ‘transparent group’ receiving original study and code or ‘opaque group’ receiving only amethod and results description and no code. The transparent group mostly verified original results(95.7% same sign and p-value cutoff), while the opaque group had less success (89.3%). Second-decimal place exact numerical reproductions were less common (76.9 and 48.1%). Qualitativeinvestigation of the workflows revealed many causes of error, including mistakes and proceduralvariations. When curating mistakes, we still find that only the transparent group was reliablysuccessful. Our findings imply a need for transparency, but also more. Institutional checks and lesssubjective difficulty for researchers ‘doing reproduction’ would help, implying a need for bettertraining. We also urge increased awareness of complexity in the research process and in ‘pushbutton’ replications.

