hello! in general, the goal of this project is to develop a simulation in the R programming statistical environment to investigate the assumptions that drive mixed-effects/random coefficients/multilevel/hierarchical linear models (so many names to describe the same thing).
in the ideal setting, the user should be able to specify:
- the number of simulation runs (as expected)
- the sample size
- the regression coefficients (so one can specify them and see if violations of the assumptions biases them)
- the intra-class or intra-cluster or intra-group correlation coefficient (again many names for the same concept)
- the probability distribution of the residuals/errors (they are always assumed to be multivariate normal. one should be ablet to change that to other skewed distributions and whatnot)
the analysis would ideally be done in the R package lme4, but if you are only familiar with nlme we can accommodate that (i need it to be R, however. that i cannot change).
my ideal candidate for this project would be a statistician/mathematician, economist or a quantitative methodologist with a very good knowledge of R, and who is also familiar with mixed-effects/random-coefficients/multilevel/heirarchical linear models. i consider myself to be an intermediate-level R user (i have ran my own simulations, albeit simpler) with a fair ammount of statistics under my belt, so i can communicate efficiently and know very clearly what i need.
a well-commented code is a *must*, since i also want to understand how things are done.
i have access to a sizeable grant so, ideally, this would be the first of a long line of simulations. in the best-case scenario i hope to find someone (or a group of people) who i can give several simulation projects to work on during the year and build a good business relationship.
this is a school assignment, so i'm keeping my fingers crossed that it shouldn't take an expert more than a week or so to finish... ;)
i sincerely appreciate your time!