Qiana Brown posted on Wednesday, June 04, 2014 - 2:07 am
Hello, I want to make sure I am interpreting the probit regression coefficients correctly in my mediation model, where I used path analysis. Prior to importing my data to Mplus I coded categorical variables using a 0, 1 convention (e.g., 0 = non-smoker, 1 = Smoker).
However, in the Mplus output file under "UNIVARIATE PROPORTIONS AND COUNTS FOR CATEGORICAL VARIABLES" the categories for my three dependent variables are described using 1s and 2s ( "category 1 & category 2") instead of 0s & 1s.
My question is, what reference groups are being used in the probit regression models for the categorical outcomes and predictors? Does the regression model use my original 0, 1 coding, with 0 as the reference group, or are 1s and 2s used, with 1 as the reference group? Thank you.