Hello,
- I've ran Multiple Linear and Logistic Regression Analysis in R in order to compare the effect of a couple of predictors including Sex and group for IQ (categorically defined as High or Average).
- I've been intrigued by results differences when using lm and glm functions when I :
1) the parameters remain coded as factors
2) the parameters are converted to numeric using as.numeric()
- considering there isn't any missing values
Hello Judicaël Fassaya. You are getting different results because R handles metric and categorical variables in different ways. Regarding regression models and concerning categorical independent variables, R tests if group "b" differs from group "a" in regard to the dependent variable. In such cases the alphabetic first group is always used as "reference". On the other hand, if your independet variables are metric R tests if the dependent variable is influenced by the independent one.
In your case I would include Sex and IQ as categorical variables in the model.
I hope this helps.
Best regards Paul
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running