# Relationship between continuous and categorical variables with missing values

I have a dataset with 2700 obs, it is the response based on SDQ questionnaire I have a categorical variable with more than one category and missing values and have some other continuous variable with missing values too. the proportion of missingness is significant and I cannot ignore them how can I analyze the association between these categorical and continuous variables?

## migrated from stackoverflow.comMay 23 '18 at 2:42

This question came from our site for professional and enthusiast programmers.

• You may need to check different imputation techniques and choose one that will be suitable for your dataset. – Confusion Matrix May 23 '18 at 2:23
• What analysis do you want to do? Regression? Making a plot? – pdb May 23 '18 at 4:31
• The mice package has functions for missing data imputation. Depending on the type of analysis you're doing, for the categorical variable you could also treat the NA's as a distinct (valid) category. – Dominic Comtois May 23 '18 at 4:33
• I have variables with response based on likert scale having 5 categories and a "I don't know" category. But these variables also have missing values. I have to analyse both the missingness and "don't know" values and perform suitable imputation for it. Also I have these data from an intervention program with 2 more variables sex and grade which are binary. I have to develop a suitable model to predict the possibility of a student being female/male and his grade given these likert variables and some other continuous variables with many missing values about the attitude of the student. – rishabh saraf May 23 '18 at 15:48

You might also want to check the book on imputation by Schafer. For mixed variables, the package mix in R implements the method.