Hello everyone,
I need some help regarding multicollinearity. I have used both Continuous and Categorical independent variables( 2 continuous and 10 categorical ) and the sample size is 392.
Model used: Binary logit
The objective of the study: What determines households enrollment in CBHI scheme.
Software: STATA package
-Can I use VIF for both Continous and Categorical variables to detect multicollinearity?
Any answer is appreciated!
Thanks
Takele,
Hello Takele Ayele. Fox & Monette (1992) describe a generalized variance inflation factor (gVIF) that can be used with categorical variables. I know there is an implementation of it in R, but am not aware of one for Stata. Here are some links.
- https://www.jstor.org/stable/2290467?seq=1#metadata_info_tab_contents
- https://socialsciences.mcmaster.ca/jfox/Courses/soc740/lecture-7-notes.pdf
Given the topic of multicollinearity, I cannot help also providing these links:
- https://statisticalhorizons.com/multicollinearity
- https://davegiles.blogspot.com/2011/09/micronumerosity.html
HTH. ;-)
You can use the command in Stata:
1. Continuous outcome:
regress y x
vif
2. Binary outcome:
logit y x, or
vif, uncentered
A rule of thumb is VIF>5 (some people say 10) poses serious multicollinearity: Article A Caution Regarding Rules of Thumb for Variance Inflation Factors
Thank you in advance Bruce Weaver & Md. Belal Hossain. You gave me a good hint and reference.
An important question that deserves to follow ... Thank you very much
- Badges
- Science topic
- Similar topics
- Psychology