Reputation: 127
I want to run a Binary Logistic Regression with the Dependent Variable separation (0 = No, 1 = Yes) and the type of relationship (also categorical with 3 categories). Now I would like to choose a different contrast definition than the one SPSS saved in advance. I would like to choose the first contrast so that only category 1 and category 3 are compared (weights: 1/2, 0, -1/2) and as second contrast the weights so that the mean of category 1 and 2 are compared with category 3 (weights: 1/3, 1/3, -2/3).
Does anyone know how I can define my own contrasts? Or is that not possible in SPSS? I would have otherwise created my own variables with the weights and added them as predictors.
Kind regards and thank you!!!
NatasaSu
Upvotes: 0
Views: 142
Reputation: 11
NatasaSu, Hi. I do not know about this specific statistic, but you can definitely do this in general. You may already know this, but you can find some nice help by doing a text search for the LMATRIX, MMATRIX, KMATRIX, or the CONTRAST subcommands in the SPSS "Command Syntax Reference." Sincerely, Dante
Upvotes: 1