0
$\begingroup$

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?

$\endgroup$

migrated from stackoverflow.com May 23 '18 at 2:42

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

  • $\begingroup$ You may need to check different imputation techniques and choose one that will be suitable for your dataset. $\endgroup$ – Confusion Matrix May 23 '18 at 2:23
  • $\begingroup$ What analysis do you want to do? Regression? Making a plot? $\endgroup$ – pdb May 23 '18 at 4:31
  • $\begingroup$ 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. $\endgroup$ – Dominic Comtois May 23 '18 at 4:33
  • $\begingroup$ 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. $\endgroup$ – rishabh saraf May 23 '18 at 15:48
0
$\begingroup$

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

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.