## Fitting nlme ## I'm not sure why I wrote this test library(nlme) set.seed(101) run.cv.response <- FALSE if(run.cv.response){ dat <- expand.grid(time = 1:10, rep = 1:4, trt = letters[1:6]) dat$eu <- with(dat, paste0(trt,"_",rep)) dat$y <- c(replicate(24, SSlogis(1:10, 10, 5, 1.5) + rnorm(10, sd = 0.25))) datG <- groupedData(y ~ time | eu, data = dat) plot(datG) fitL <- nlsList(SSlogis, datG) fnm1 <- nlme(fitL, random = pdDiag(Asym + xmid + scal ~ 1)) fxf <- fixef(fnm1) fnm2 <- update(fnm1, fixed = Asym + xmid + scal ~ trt, start = c(fxf[1], rep(0, 5), fxf[2], rep(0, 5), fxf[3], rep(0, 5))) anova(fnm2) plot(fnm2) }