Is it possible to preregister non-experiment studies?

Question

I am a PhD student in social science and a professor and I are planning to run a survey in South Asia. It is not an experiment. However, we were thinking of pre-registering in case it would give us more credibility when we submit it to journals. Do places like OSF accept preregistration for nonexperiment studies?

Answer 1

It is probably worth stepping back a bit here and recalling the scientific purpose of pre-registration. One of the main goals of pre-registration is to avoid situations where researchers change their hypotheses and methods post hoc after they have already seen the data from their study (or simply don't publish their results as a post hoc decision). This is known to lead to significant confirmatory bias in science, so we want to try to avoid this happening. Pre-registration works by having the researchers specify their research plan, including the hypotheses they will test and the methods, models, etc., they will use for their inferences. The pre-registered explanation of these things allows readers to check to see if there has been any funny-business in the testing that ultimately occurred in the research (e.g., post hoc testing of hypotheses arising from looking at the data), and it allows them to see if there have been studies that were planned/executed but not published (to mitigate the file-drawer problem). Pre-registration serves this purpose irrespective of whether a study is based on controlled interventions and resulting causal inferences (e.g., in a randomised controlled trial) or merely passive observation and resulting predictive inferences (e.g., from a survey).

The Open Science Framework (OSF) "...welcomes preregistrations from any field or discipline and we support researchers using OSF as a preregistration platform whether or not they will also be using its data storage, collaboration management, or publishing capabilities". You can create a pre-registration of planned research by filling out the available form, which usually entails specifying some metadata for your study and basic study information, your design plan, sampling plan and analysis plan (including specification of the variables you will collect and examine) and any other relevant information in your research plan. You can then refer to your pre-registration in your published research to show the reader that these aspects of your research plan were pre-specified publicly prior to collecting the data in your study. (And of course, you should tell the reader if you have departed from your pre-registered plan in some way.)

So, in answer to your question, yes you can pre-register a "non-experimental" study (by which I presume you mean one without any intervention present). Pre-registration is still useful in this case and it will give your research more credibility. The reason it is likely to have more credibility is because readers will be able to confirm that you did not make post hoc changes to your hypotheses and/or methods after seeing the data from your survey.