## Problems with pooling Multiply Imuputed datasets, of multilevel logistic models, using R (MICE) - missing coefficent

Welcome to the forum for R2MLwiN users. Feel free to post your question about R2MLwiN here. The Centre for Multilevel Modelling take no responsibility for the accuracy of these posts, we are unable to monitor them closely. Do go ahead and post your question and thank you in advance if you find the time to post any answers!

Go to R2MLwiN: Running MLwiN from within R >> http://www.bris.ac.uk/cmm/software/r2mlwin/
divyanair
Posts: 1
Joined: Fri Jul 13, 2018 6:45 am

### Problems with pooling Multiply Imuputed datasets, of multilevel logistic models, using R (MICE) - missing coefficent

hello
I am having problems with the MICE package in R, particularity with pooling the imputed data sets.

I am running a multilevel binomial logistic regression, with Level1 - topic (participant response to 10 questions on different topics, e.g. Darkness, Day) nested within Level2 - individuals.

The model is created using R2MLwiN, the formula is > fit1 <-runMLwiN( c(probit(T_Darkness, cons), probit(T_Day, cons), probit(T_Light, cons), probit(T_Night, cons), probit(T_Rain, cons), probit(T_Rainbows, cons), probit(T_Snow, cons), probit(T_Storms, cons), probit(T_Waterfalls, cons), probit(T_Waves, cons)) ~ 1, D=c("Mixed", "Binomial", "Binomial","Binomial","Binomial", "Binomial", "Binomial", "Binomial", "Binomial", "Binomial" ,"Binomial"), estoptions = list(EstM = 0), data=data)

Unfortunately, there is missing data in all of the Level1 (topic) responses. I have been using the mice package ([CRAN][1]) to multiply impute the missing values.

I can fit the model to the imputed datasets, using the formula > fitMI <- (with(MI.Data, runMLwiN( c(probit(T_Darkness, cons), probit(T_Day, cons), probit(T_Light, cons), probit(T_Night, cons), probit(T_Rain, cons), probit(T_Rainbows, cons), probit(T_Snow, cons), probit(T_Storms, cons), probit(T_Waterfalls, cons), probit(T_Waves, cons)) ~ 1, D=c("Mixed", "Binomial", "Binomial","Binomial","Binomial", "Binomial", "Binomial", "Binomial", "Binomial", "Binomial" ,"Binomial"), estoptions = list(EstM = 0), data=data)))

However, when I come to pool the analyses with the call code > pool(fitMI) it fails, with the Error:

Error in pool(with(tempData, runMLwiN(c(probit(T_Darkness, cons), probit(T_Day, :
Object has no coef() method.

I am not sure why it is saying there is no coefficient, as the analyses of the individual MI datasets provide both fixed parts (coefficients) and random parts (covariances)

Any help with what is going wrong would be much appreciated.

I should warn you that this is my first foray into using R and multilevel modelling. Also I know there is a MlwiN package ([REALCOM][2]) that can do this but I don't have the background to use the MLwiN software.
Minecraft Pocket Edition Counter-Strike Google Play Services