I'm following the suggestion from this post https://www.cmm.bristol.ac.uk/forum/vie ... a976#p4572 to replicate the standardised residual vs predicted value within R.
I am two models. Model 1 is a 2-level random slope model with continuous outcome and IGLS is used to estimate the model. Model 2 is a 2-level random coefficient cumulative logit model with a ordinal outcome with 3 categories and MCMC estimation is used.
I have no problem to calculate the predicted value for Model 1 using the following code:
Code: Select all
pred <- data.frame(level2 = model1@data$lv2, pred.outcome = predict(model1))
pred.outcome.mean <- summaryBy(pred.outcome ~ level2, data = pred, FUN = mean)$pred.outcome.mean
Code: Select all
pred <- data.frame(level2 = model2@data$lv2, pred.outcome = predict(model2))
I've also tried to add type = "response" in the predict(), the same error appeared.Error in `[.data.frame`(indata, x.names) : undefined columns selected
Anyone has any idea?
Thanks a lot.
Vivian