Hi.
I am trying to analyze moderating effect in HLM 8.
And I also input control variable.
But the problem is control variable has "and so on" option.
So, when running analysis, I find some of my data got loss.
Problem related control variable is position and supervisor position.
These variables have 1 to 5 point. (1. staff 2. administrative manager 3. section chief 4. chief 5. manager)
In this situation, how can I proceed analysis without data loss?
I remember but not exactly that in some of study, "and so on" option became average point
throughout control variable. (Ex. "and so on" become 2.3, because average position in the study is at 2.3).
Is it possible or are there another alternative?
and so on from what ? It does not mean much to me. Could you provide a tabulation of the problematic variables which would help us. I presume that you can going to use some sort of dummy coding so you could just create another category, but I really do not understand what you are trying to do.
Have a look at https://stats.idre.ucla.edu/spss/faq/coding-systems-for-categorical-variables-in-regression-analysis-2/
Kelvyn Jones What I tried to do is 'Mean imputation'
I found that there is multiple imputation method but It was hard for me.
So I tried just putting missing data with averaged score(position and supervisor position) respectively.
Is it okay to do mean imputation and analyze my data?
If you have a categorical variable I see no point in doing mean imputation. Can you explain the problematic variable in more detail?
- Badges
- Science topic
- Similar topics
- Psychology
- Organizational Psychology