datsim: Repeated measures for two groups of subjects drawn from mmcar model with no nuisance covariates
Description
A simulation dataset containing repeated subject measures for 2 treatment groups,
(control = 0, treatment = 1), constructed from an mmcar model with correlation
between adjacent sessions equal to 0.25. Subject effects were randomly drawn from 10 clusters
with weights/probabilities drawn from a Dirichlet distribution. Cluster location
values were generated from a Gaussian base distribution.
Format
A list object of 19 variables for 792 total observations on 264 subjectsDetails
- y. response. there are
N = 792 total measures for P = 264 subjects
- subject. subject identifier
(1,2,...,264
- trt. treatment group identifier of length
N (e.g. (0,0,0,...,1,1,1,...) , either {0,1} for control and treatment.
- time. times in months for each repeated subject measure of length
N. There are 3 distinct time points. e.g. (0,3,6,0,3,6,0,0,3,,,,)
- n.random. number of random effects per subject. Set = 3.
- n.fix_degree. order of fixed effects. Set = 2, for quadratic, meaning 3 effects (intercept, slope, quadratic) each, for treatment and control groups.
- coefs. true fixed effect coefficient values used to generate data.
- subj.aff. indexes subjects receiving treatment.
- W.subj.aff. multiple membership weight matrix that maps the
P_aff = 132 affected subjects (in subj.aff) to any of S = 245 treatment sessions.
- group. treatment group membership for each of the
S sessions.
- Omega. the
S x S CAR adjacency matrix used to model prior dependence among sessions
- gamma. true session effect values (of length
S) used to generate model response.
- s. true cluster memberships for each of the
P subjects.
- b.star. a list object of true cluster location values for each of
M = 10 clusters. Each entry contains the n.random = 3 location values for that cluster.
- b. a list object true random effect coefficient values for each of
P subjects. Each entry contains the n.random = 3 effect values for that subject.
- tau.b. true values for the prior precisions of the base Gaussian distribution for each of
n.random = 3 subject effects.
- tau.e. true value for overall model error.
- coefs. true coefficient values for the time-based quadratic fixed effects generated from the
trt, and time inputs.
e.g. X = c(1,time,time^2,trt_1,trt_1*time,trt_1*time^2).