Dear all,
I wonder if anyone has any views on this issue?
Say, hypothetically, we were doing a binary logistic regression and we had a set of categorical variables each on a 5 level likert scale .... 1=strongly disagree....5=strongly agree. For some of these variables there are very few subjects whose responses are in the "extreme" category i.e. for a specific question there could be <5 subjects who strongly disagree. Say we intend to enter each of these q level 'likert scale' questions into an analysis as q-1 dummy binary variables. Are there any rules/criteria which say something like "a category which has fewer than x responses" should be considered for merging with an adjacent category?
Having said this, I know entering likert scale variables into a model in their ordinal form is controversial, but if we *did* enter them as ordinal variables....I suppose the "merging" issue would not come into play i.e. it would not matter if, for some of the variables, there are very few subjects whose responses are in the "extreme" category(ies)?
Many thanks for your views in advance.
Kind Regards,
Kim
Dr Kim Pearce PhD, CStat, Fellow HEA
Senior Statistician
Faculty of Medical Sciences Graduate School
Room 3.14
3rd Floor
Ridley Building 1
Newcastle University
Queen Victoria Road
Newcastle Upon Tyne
NE1 7RU
Tel: (0044) (0)191 208 8142
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|