I have a list of 15 traits that subjects can ascribe to a topic. The selected traits then get carried forward to a scale where they rate the perceived strength of each trait. The selected traits could range from none to all 15 for a subject. For e.g. subject 1 could have 3 selections, subject two could have 11 selections, and so on....
How can this type of data be treated as IV's?
For this type of data in which questions having multiple choice answers i.e the answer may be 1 or 2 or even all. Then we have to make variable for every option, then it should be treated as IV's. Suppose a question have 4 options, then for that question there should be 4 IV's according to option.
Another method is that you should assigned some weight to the every option and then combine it when doing analysis.
Dear Sumiran,
I think you will need to include all 15 traits as variables in your model, with a score of 0 representing not present. The issue is whether people who choose more traits should be allowed to score them equally to those who select fewer. You could use a percentage weighting to make it fairer. I'm not sure on the statistical validity of doing this, but scoring as an ordinal value for each trait is also problematic as this is not scale data.
- Badges
- Science topic