First, I'd check if the data set for the study is available in an online repository (e.g. OSF). If that doesn't work, you could reach out to the authors and ask if they'd be willing to share their data set with you. If this fails, you can use simulated data to create the model and conduct a power analysis on the simulated data using simr. You'll just need to know the slopes and intercepts for the fixed and random factors (which the author should have included in the results of the published work.). Even better: run analyses on the simulated model multiple times adjusting the parameter values each time to see how different values influence power and required sample size.

Hello Claire,

I am far from having expertise in statistics but I doubt that this is reasonable as the prior estimate is still just an estimate in a unknown sampling distribution. As far as I know power calculations (for which simulations of your model are the way to go as Blaine correctly said) should be based on your theoretical assumption of the population parameter.

Just my 2c

--Holger

Dziak, J. J., Dierker, L. C., & Abar, B. (2018). The interpretation of statistical power after the data have been gathered. Current Psychology, 1-8. doi:10.1007/s12144-018-0018-1

Arend, M. G., & Schäfer, T. (2019). Statistical power in two-level models: A tutorial based on Monte Carlo simulation. Psychological Methods, 24(1), 1.

Thank you Claire Linares for your question. Try this link, it explain the steps

http://docs.wingarc.com.au/superstar/9.10/superchannel/reference/loading-spss-survey-data/example-of-channelling-spss-survey-data

Also, Claire Linares I suggest looking at this resource. If still not sure contact IBM the provider of SPSS to help you. They will. Good luck.

https://www.ibm.com/support/knowledgecenter/en/search/open%20SXV4%20on%20SPSS?scope=SSLVMB

Worth looking at http://www.bristol.ac.uk/cmm/software/mlpowsim/. As far as using previous results, it is better to decide what is the minimum effect size that you want to design your study to detect with whatever likelihood.

Daniel Wright ; the mlpowsim software is indeed designed to do what is required ; it also has a very useful and readable manual that covers the issues.