Utilising single-subject parameter distribution information in JASP/R
Hello,
I am using JAGS/winBUGS to estimate two parameters of a model for each participant, let's call them xi and alpha. I have two independent samples (groups) of uneven size. I would like to perform a Bayesian t-test - that can be done in JASP. However, I would like to utilise the information about the certainty (parameter distribution) of each parameter estimate (given by the posterior distribution). I want this because i noticed that xi has very low standard deviation in all subjects (0.0002) while alpha has very high standard deviation (0.15). If I use the point estimate of the mean and perform a Bayesian t-test both come out significant but when I look at the certainty it really shouldn't in the case of alpha.
1) Is there a way that I can incorporate this information in my test?
2) Could I use the same information to perform a mixed ANOVA?
3) If JASP is not there yet, is there any R/Matlab package that might have already developed this?
Thank you!
Ondrej
Comments
Hi Ondrej,
Great question. Ideally, you'd integrate the ANOVA structure with the model underneath. When that is difficult some two-step method could be used. Here's a paper that may be relevant: https://www.collabra.org/article/10.1525/collabra.78/
Cheers,
E.J.
Hi EJ,
thank you
I will have a go.
Ondrej