Abstract

The proposed symposium consists of three presentations and a conclusion.

The first presentation is entitled “DOES FAILURE TO REPLICATE AFFECT CITATIONS?” It reports the results of an analysis of 225 replicated studies to determine whether unfavourable replications adversely affected their citations. The study consisted of two parts. First the replications were assessed to determine whether they “confirmed”, “refuted”, or gave a “mixed report” about the original study. Each original study was then compared to a pool of over 100,000 other studies to find matches that had identical or similar citation records at the time the original study was replicated. We then compared the citation records of the replicated and matched, unreplicated studies in the post-replication period.

The second presentation is entitled “PAVING THE ROAD TO REPLICATIONS: RESULTS OF AN ONLINE EXPERIMENT”. This presentation reports the results of a 12-month, online experiment in which researchers were alerted to the existence of a replication while searching for studies. The online experiment tracked whether researchers were more likely to check out the replication after being alerted to its existence. Further details cannot be given at this time because this experiment is still live. It will end in July 2021, at which time details about the experiment can be made public.

The third presentation is entitled “”Feasibility of Replication Studies in Behavioral Ecology””. Most ecologists believe close replication studies are important, yet they do not conduct or publish them (<1% of published studies in ecology claim to be replications). An often-cited reason for the disconnect between researchers’ beliefs and actions around replicating original studies is the difficulty of reproducing ecological conditions, making replications infeasible. The presentation explores the prevalence and feasibility of replications in behavioral ecology, a curiosity-driven subfield of ecology that seeks to understand why individuals behave the way they do. After presenting results on five components of replication feasibility (accessibility of the system, intensity of work, duration of work, complexity of the work, and how expensive the work would be), this presentation explores why replication rates remain low even when feasibility is high.

The symposium will conclude by discussing the implications of this research for the effectiveness of replications to provide a disciplinary device that encourages scientific integrity.