Dear all
I would be very interested to receive constructive suggestions (particularly rules of thumb where possible) as to the stage at which a dependent variable may be regarded (albeit crudely) as continuous. To explain further, I am specifically interested in a scenario where the dependent variable ranges over values 1, 2, 3, ... n, where n is some fixed value and the question is which factors are associated with the dependent variable. In such cases, I might for example, choose to transform the dependent variable into a binary variable by categorization and perform a binary logistic regression analysis. However, I might also wish to use the raw data and carry out a multiple regression analysis (provided the necessary assumptions are satisfied). My more specific question therefore could be presented as, what value of n is high enough in order for me to consider my dependent variable as worthy of consideration for a multiple regression analysis (ignoring for now all other issues !
re
assumptions to be tested)?
I look forward to receiving your views
Best wishes
Margaret
---------------------------------
Yahoo! Messenger NEW - crystal clear PC to PC calling worldwide with voicemail
|