In its core, the CATS project developed a solution for secure, transparent, collaborative work on shared, privacy-sensitive data to increase replicability of research findings.
The CATS project started in 2015, and it was set up as a collaborative, open project in all steps. To credit original authors for their work on collecting the data, we invited them to become ‘collaborators’ on the project, making them part of the whole study process. Currently, the CATS project consists of over 70 collaborators. We arranged a Collaborators Meeting as a preconference to a key conference in our field, during which we discussed which research questions to study and how we would pursue them. Writing articles is also a collaborative effort: anyone can pitch a research question, pitches of articles are circulated across the entire CATS group, and everyone can indicate interest to be actively involved in writing the article.
Collaborating in a responsible manner with so many authors was challenging, because we had to protect the privacy of participants while also receiving the data from across the globe. For this, we had endless discussions with the privacy officers to come up with GDPR-compliant data sharing agreements. Together with the Research IT team, we used VU’s unique SciCloud-solution to build a data commons for storage and analysis, which is a secured, remote-access information-technology infrastructure holding the pooled data set, syntax codes, and various analysis software packages. Pseudonymized data are shared through the secure environment of Surfdrive, after which the data cleaning takes place on SciCloud’s virtual machine. Finally, the merged dataset is placed on the data commons. The carefully crafted data sharing agreements in combination with the data commons on SciCloud made our vision for robust, collaborative work in this field a reality.
Expanding the project
After the first years of the project and the first presentations of the study methods at conferences, other researchers approached us telling us they had ‘caught the bug’ and that they wanted to set up IPD meta-analyses in related areas of our field. The CATS project was broadened to CARS to encompass the new spin-off projects. Together with several collaborators, IPD efforts into Attachment Interventions (CAPIS) and Attachment with Multiple Parents (CAMPOS) were started. These spin-offs are also running on our Data Commons platform.
Sharing our approach
We recently published our experiences with this way of data pooling and large-group collaboration in Current Directions in Psychological Science (Verhage et al., 2020). We were also invited for international key note talks on this topic. More practically, we’ve shared our research protocols, presentations, and data templates on our OSF page, as part of a reproducible workflow. We are still learning to work more ‘open’ as well. For CATS, we placed the protocol of the study on the OSF, but with later projects, we routinely pre-register study hypotheses and protocols.
Our next step will be to develop the data commons as an open data facility, where researchers can run analyses through an interface that keeps the data itself hidden.
Read the full Open Science>