Missing Data/Imputation Discussion > MI with non-normal ordinal data
Great question, Aaron!
Are you looking to impute likert items that are part of a scale (that sums the items), or are the items themselves part of the analysis?
March 22, 2012 |
Jeremy Taylor
Jeremy Taylor
Greetings, I am wondering whether anyone is familiar with a way to perform multiple imputation analysis with variables that are Likert (ordinal) scale and not multivariate distributed. The thought of dichotomizing the variable is concerning, because of the quality of data that would be lost, so I would like to know there is a way to handle this problem or if it would just be best to dichotomize?
Thanks in advance!