LUCS Seminar: Andrey Anikin - Can I trust this paper? Replication crisis and other bugs in quantitative research

We all regularly read papers, perform experiments, analyze our data, and write more papers in which we cite previous research and tell the world about our own awesome findings. Is there a problem with this time-honored life cycle of empirical research? In this round-table discussion, we will attempt to summarize a decade of “replication crisis” in psychology and other empirical disciplines. The key takeaway is (probably) not to trust all published data blindly, but to watch out for serious methodological and conceptual problems such as data falsification, lack of reproducibility of reported analyses, replication failures, over-generalization, poorly specified hypotheses, and structural shortcomings within the publishing industry. It’s not all doom and gloom, though, as many of these problems can be addressed if we become more aware of them both when reading other papers and when publishing our own.