[R-sig-ME] Simulated mean prediction (lmer or lme) and CI
brook@@log@n @ending from gene@com
Thu Nov 15 17:38:27 CET 2018
I'm hoping to get some guidance on doing double delta cQTc analysis using
linear mixed effects. I have not done this type of modeling before but I
want to create a single fitted line for my data with the associated CI and
then calc the difference between my TX arms (0,1 binary treatment
indicator) and the CI on those estimates but I'm stumped on whether I'm
calculating things correctly or not.
Below is my current model
check3 <- lmer(CHblQTCF ~ TX + CP + popDQTcF + TIME1 + (1 +
CP|Subject),REML = FALSE, data = eg_final)
(using merTools:: predict interval)
PI <- predictInterval(merMod = check2, newdata = eg_final,
level = 0.90, n.sims = 1000,
stat = "mean", type="linear.prediction",
include.resid.var = TRUE)
This gives individual fits and CIs for each subject but how would I get a
mean fit and CI for the population?
I believe something along these lines is required.
But I don't know what I would need to modify for a linear model.
Thanks for you assistance,
[[alternative HTML version deleted]]
More information about the R-sig-mixed-models