I am using the MICE package to do multiple imputation. Once I generated my imputed datasets, I wanted to make certain variables binary and certain variables factors. I figured out how to convert my mids file to a dataframe and edit it using the "complete" function, but evidently, the "pool" function doesn't work with a data frame. How can I make these changes to variables in a mids object? Or, figure out a way to get a single estimate (similar to the pool function) from a dataframe? Since I don't need these recoded and averaged variables for imputation, I would rather avoid passive imputation, if at all possible. Thank you, Joanna [[alternative HTML version deleted]]