Summary of the interview
It seems that only 10-40% of research can be replicated. While 100% is unrealistic, there is room for improvement. The prevalence of replication issues has been well reported in some research fields but we do not have a complete picture on all fields yet. In principle, everywhere where data is used, it can happen.
Underlying causes of the problem
There are several causes that contribute to the problem: small sample sizes, publication bias, p-hacking and the research culture such as pressure to publish. Some disciplines have already learnt from their mistakes. Animal research is a good example where the problem was recognized some years ago and where there are now better guidelines in place to make research more reproducible. Researchers are often unaware of the problem. They do not realize that they have to first, clearly define the research question, pre-register the full study protocol and then execute the research. Furthermore, the details of the methods have to be reported well so that others do have the necessary information to redo the study.
Replication research has a reputation of not being “sexy science”. That seems to be wrong. There has already been two calls by the Netherlands Organisation for Scientific Research (NWO) for replication studies. The proposals have to fulfil some requirements. For instance, the research question must be relevant and the study was not replicated before but can be replicated (i.e. the necessary information on how it was done is available). The researchers who received the funding were often members of the Young Academy (KNAW), the ’rising stars’ of science in the Netherlands. They obviously see the importance of replication. Ideally, in some years it should be normal to replicate studies and special calls are less necessary than now. The most important aspect is that science becomes more transparent on the research plan and methods. The problem of replication crisis might then disappear.
Summarized by Claudia Lüttke