Hi, I have a dataset with clustered data (observations within groups) and would like to make some descriptive plots.
Now, I am a little bit lost on how to present the dispersion of the data (what kind of residuals to plot). I could compute the standard error of the mean (SEM) ignoring the clustering (very low values and misleading) or I could first aggregate the data by calculating th mean for each group and calculate the SEM for this means. But I am not so sure what implication these two approaches have. In the end, I take the clustering into account by fitting a random-intercept regression model – however for plotting I would like to have a descriptive dispersion indicator of the data. Now, I heard a lot about 'clustered' or 'robust' standard errors. Is there some kind of correction I can apply to the simple SEM formula (sd(x)/sqrt(m)) to take care of correlated observations within clusters? Or are there bootstrapping or jackknife approaches implemented in R (or cran package) which give me unbiased variance estimation for clustered data? thanks for any suggestions! ______________________________________________ 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.
