Preregistration of research provides greater transparency

YOUth is a large scale, longitudinal research project on child development. Almost 10 years ago, Chantal Kemner received a gravity grant to set up the YOUth research within a national consortium. From the start, YOUth is all about Open Science. All data collected through YOUth are available for other researchers, so-called FAIR and open data. But being open is not something you just do. It is a long road that is be done step by step. One of those steps is pre-registration, it facilitates the verifiability of research results, and the discoverability of research projects using YOUth data.

Since 1 March, researchers who will use YOUth data have been sharing their precise plans with the scientific community before even analysing the data. They are doing so on the Open Science Framework, an international platform of the Center for Open Science (COS). This working method, known as preregistration, fits in perfectly with the UU's Sharing Science ambitions. Preregistration improves the reliability and transparency of the research, explains project manager Coosje Veldkamp.

Through the YOUth cohort study, Utrecht University and UMC Utrecht are collecting a wealth of data on child development. From foetus to teenager and from cerebral to behavioural development. ‘If you are all investing so much in good, large-scale data collection, you naturally want this data to be used for as long and as effectively as possible, also by researchers from outside Utrecht’, says Coosje Veldkamp. ‘That's why Open Science has been a spearhead since YOUth was first launched (in 2013). Preregistration fits in with this. If you share your hypothesis (or hypotheses) and analysis plans in advance, this helps to protect you from the unintentional misinterpretation of your analyses. In addition, it allows academics to take a critical look at the correctness of the planned analyses and check whether the published results tie in with this.’

A place of our own for YOUth research

Another advantage of preregistration: researchers can give each other tips or see if they can work together. Worldwide. ‘We have recently gained our own place (a branded registry) on the Open Science Platform. This makes studies conducted with our data easier to find within the entire academic community. It also benefits the findability and therefore the durability of our (secure) data.’

So how does it work exactly, preregistration? ‘We have revised the procedure for researchers to request YOUth data. Applications can now be made in an online system, in which we have immediately included preregistration.’ Coosje (who did her own doctoral research on preregistration): ‘It does mean a bit more work at the front end of the research process, but ultimately it saves you time: once you have completed the preregistration, you will already have completed the introduction and the method section of your research paper.’

Protection against the possibilities 

Coosje assumes that preregistration will help researchers. ‘It protects you from the possibilities offered by inferential statistics.’ How? It all comes down to the difference between testing a hypothesis or generating a hypothesis. Coosje Veldkamp: ‘You have to indicate at preregistration: we expect this and that relationship between this and that and are testing this by means of this analysis in that group. If you formulate your research question too broadly, and measure and analyse everything that is available about it in the data set, you are bound to find a number of relationships. However, we know that with the most commonly used statistical method (‘Null Hypothesis Significance Testing’), the probability that a found relationship is based on chance is already 5%, and this probability only increases if you do multiple analyses

It is therefore better to formulate precisely in advance which variables you will be using and which relationship you expect to find. If you happen to find a different relationship in additional analyses, you will need to regard this result as a new hypothesis that emerges from this study. This will then need to be investigated in more detail, with a different data set.’ The fact that you can inspire other researchers to carry out research into this new hypothesis is therefore an additional advantage of preregistration. 

Who is it for?

Preregistration is also relevant for researchers who do not test hypotheses on the basis of the most common statistics. ‘For example, data scientists often work with completely different models. It is also good to share your plans in this case, so that other academics can provide you with constructive feedback. In addition, preregistration allows you to contribute to a repository that provides a good overview of all the research that we conduct.’