In human words, we the random effects can be described as follows: We believe that the variability comes from each subject behaving uniquely at each period and from each eye of each subject at each period behaving uniquely.
As far as the crash, I've compiled a structurally-similar dataset (without all the names, etc.) and made fake data for the response variable. So please find the file attached. This should be reported to the R developers, I supose. http://r.789695.n4.nabble.com/file/n3297130/crash_data_3.csv crash_data_3.csv The model called: fit = lmer(resp ~ (subject + period + treatment):eye + (eye|subject:period), data=sub) Hi dunner, I got a pdf of the book, thank you for the advice. I will take me a while to digest all the info in there. ... I think I managed to get it right. I used the "lme" function from the "nlme" package. I've combined the subject and period variables into a new variable called "SP" and used the call lme(resp ~ (subject + period + treatment)*eye, random=~ 1|SP/eye, data = sub) Which explots the fact that a "/" in the random effects will take the entries separated by it in heirarchical order and their combinations. So it gives me "SP" and "eye %in% SP"... the anova table and the summary seem to be correct. I just need my supervisor to check the result and then I can move on. Thanks for all the help. :) -- View this message in context: http://r.789695.n4.nabble.com/Unbalanced-Mixed-Linear-Models-With-Nested-Stratum-tp3263969p3297130.html Sent from the R help mailing list archive at Nabble.com. ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
