Hi - Thanks for the tutorial, very useful. However, I am having trouble in that SPSS is only replacing values for my quantitative variables, and not for my categorical ones. Even the help screen on SPSS says that only quantitative variables can be imputed! I wondered whether it's because the Missing Values Analysis requests replacement of missing values based on a normal distribution. Some of my categorical variables have large skews which is not unexpected, but might this be why SPSS can't/won't impute the missing values for these variables? If so, do you have any guidance as to how to get missing values for these? many thanks
+Katie Rapson Are you using numeric codes for your categorical outcomes? If not, you should. If you are, then change the variable in the Variable view box to nominal.
I was trying to use MI but It is too complicated for my data and my analysis. I am not sure I can use EM because you said it is not good to be used for inferential analysis!!! Am I right? I have 73% missing in variables,45.8% in cases, and only 0.7% missing values. Could you please give me a suggestion? I really don't want to get a biased results because of using an inappropriate technique to deal with my missing values? many thanks.
Hi - Thanks for the tutorial, very useful. However, I am having trouble in that SPSS is only replacing values for my quantitative variables, and not for my categorical ones. Even the help screen on SPSS says that only quantitative variables can be imputed! I wondered whether it's because the Missing Values Analysis requests replacement of missing values based on a normal distribution. Some of my categorical variables have large skews which is not unexpected, but might this be why SPSS can't/won't impute the missing values for these variables? If so, do you have any guidance as to how to get missing values for these? many thanks
+Katie Rapson Are you using numeric codes for your categorical outcomes? If not, you should. If you are, then change the variable in the Variable view box to nominal.
I was trying to use MI but It is too complicated for my data and my analysis. I am not sure I can use EM because you said it is not good to be used for inferential analysis!!! Am I right? I have 73% missing in variables,45.8% in cases, and only 0.7% missing values. Could you please give me a suggestion? I really don't want to get a biased results because of using an inappropriate technique to deal with my missing values? many thanks.