How did you get into open science?

Joachim Hüffmeier: I come from the field of social psychology, where I did my doctorate. On the one hand, the replication crisis since 2012 has made this discipline a prominent example of the problems associated with non-open data and non-open science. On the other hand, social psychology is also a poster child for innovation in the direction of open science. Even during my doctorate, I shared the concern about "questionable research practices", i.e. research practices such as formulating hypotheses after data analyses had already been carried out. When it became clear how big the replication problems are in social psychology, I thought I would do things differently with my research group. Since 2016, all studies that start here have been preregistered as a matter of principle and we also share all data that can be meaningfully shared. We also make all information on the procedure available. We want our research to at least be replicable and have already conducted our own research on the question of what good open science looks like and how we can get more scientists to apply open science practices.

Ann-Kathrin Torka: My Bachelor's and Master's theses here in the working group have already been preregistered. The supervisors explained to us what it was good for and we've been doing it right from the start, so I don't know any different. During my doctorate, I then started not only using and applying open science practices, but also researching them.

Is data often shared in psychology?

Hüffmeier: No, it is not shared very often. However, awareness of the problem is high. The big bang that occurred here in 2012 regarding the non-replicability of research findings in social psychology will also happen in many other disciplines. In other words, we are one step further, but unfortunately it is not yet the standard for data to be shared.

Torka: We are not yet as far as we would like to be. Among other things, we have evaluated the extent to which journals require open science practices on their homepages. A third of the journals in the field of work and organizational psychology, for example, do not yet provide any information about sharing data. However, the journals have a lot of influence in this regard: some journals, for example, provide corresponding articles with open science badges. Studies have shown that authors are more willing to share their data and prepare it better if the articles are then marked with such a badge.

Hüffmeier : Unfortunately, a "data availability statement", i.e. a note in the article that data will be shared on request, does not help. There are studies that show that nine out of ten researchers do not share the data despite the statement. It is the responsibility of the journals to ensure that authors share their data by default, unless there are good reasons not to, for example data protection concerns. As long as journals do not do this, it is a clear indicator that we do not have enough awareness of the problem in this area.

What advice would you give to other scientists who are interested in open science?

Torka: It can be very helpful to exchange ideas with others who are already using open science themselves. You need a bit of time at the beginning to get to grips with it, but ultimately it's not difficult, anyone can learn it. You should take the time for Open Science because it is an investment in the future.

Hüffmeier: You don't have to use all the elements of Open Science at once. It's enough to start with preregistration, for example, and perhaps not share the data at first if you don't feel safe; or just share the data without preregistering. Committing to open science does not have to be an all-round effort that costs an infinite amount of time. Small steps also lead to the goal. Open science is not black or white - it can be gray at the beginning. The benefits for research are great because data analyses can be reproduced in this way and findings can be replicated. But researchers can also benefit individually: In the review process, both preregistrations and shared data are actually always rated positively by reviewers.