Help support open science today.
Donate Now

10 Tips for Making a Great Preregistration

October 10th, 2017,
Posted in: Preregistration

[UPDATED] Preregistration is the first line of defense we can use to protect ourselves from biases and questionable research practices. Preregistration does require some additional work ahead of time, but since it provides a clear plan of what you set out to do, the research is easier to do and the write-up goes MUCH faster. 

Since you might be new to the process, we've pondered your plight and have come up with 10 helpful tips to help you create a successful Prereg Challenge submission.

1.     Clearly describe your hypotheses, variables, and analyses. Make sure the variables you plan to manipulate and measure appear in your hypotheses and analyses so it is clear what hypothesis is going to be analyzed by a specific analysis plan.

2.     For abstract or subjective measures, describe how the variable will be measured, scored, or recorded. If an index of multiple responses is going to be created, state how this will be done.

3.     If you are using a confirmatory factor analysis to create a variable of interest, specify the number and structure of the factors (e.g. the items that load onto each factor and rotation of factors).

4.     Specify the number of tails your analyses will include. Preregistration gives you a great rationale for 1-tailed tests if you have a predicted direction.

5.     Specify the significance threshold you will use for making inferences based off your statistical tests.

6.     Make sure each hypothesis has a specific analysis plan, using the same variables mentioned in the hypothesis section.

7.     For follow-up analyses, list any interactions that you plan on testing. If you plan to investigate every possible interaction, specify this and state the total number of follow-ups you plan to conduct.

8.     Label any analysis that you cannot fully pre-specify as exploratory or preliminary.

9.     When preregistering empirical work, make sure the preregistration has at least one inferential analysis that directly tests one of the listed hypotheses. It should be very easy to tell exactly how many tests you are going to conduct. Number them if there are more than one!

10.  Fully specify the state of the data that will be used in the preregistered experiments. If you will be combining some analyzed data with the yet-to-be-analyzed data, list these analyses as exploratory, as the results of the first set of experiments have likely informed the current analyses.

Happy Preregistering! 


Recent Posts