I have the data drew from the questionnaire which is consists on socio economic characteristics, binary questions, and multiple questions, rating scale, likert questions and likert scales consists on 36 items, 8 items, 7 items, 5 items and open ended questions too. All variables are categorical. Number of respondents are 450. I have 10% missing data .
My question is should I impute all these variables which are of socio economic characteristics as some of them have no missing values should. As I have learned that the variables that have to be analysed must be in imputation model. As I have to apply test for the analysis on variables( checking the hypotheses). And secondly, can I test the relationship between for e.g gender with the behaviour I am studying in the research from likert scale items. In doing so I have to make a model in which all my variables must be there. So my question is Should one have to impute the scales separately and merge it. or include it. and can we include the necessary variables on which we have to do the tests ? and separately analyse those on which we are not applying test. Is there any possible solution because I am confused due to scales as well as my variables are too many. I dont want to delete my observations when only few answers are missing.
Please consult:
Paul Connolly, (2007).Quantitative Data Analysis in Education: A Critical Introduction Using SPSS.
- Badges
- Science topic
- Similar topics
- Psychology
- Mental Processes
- Learning