My research is related to deforestation. I want to perform a binominal logistic regression test where, the dependent variable is deforestation of raster image cells (Yes/No). How can I test the multicollinearity (calculate VIF value) among the qualitative variables? For example I want to determine the multicollinearity level among deforestation (yes/no), type of resident (legal/illegal) and income level (high/medium/low). I am using SPSS software to analyze my data set. Answers regrading the analysis with SPSS is highly appreciated. Thanks in advance.
Hi S M Asik Ullah
You can test Multicollinearit using Regression analysis and caculating VIF.
If VIF is greater than 10 then there a multicollineariny among your explanatory variables, and you need to solve this problem then regress your model again.
I hope if my response be useful for you.
For perfect Multicollinearity, that is, one independent variable is a linear combination of others, it is a linear algebra problem. It is not full ranked. In SAS, Proc logistic will test for you.
For near Multicollinearity, you use VIF to test it. For instance, in SAS, you call proc reg with option VIF. Say, if you have 5 independent variables x1, ..., x5, you do
proc reg data= dt; model y = x1 x2 x3 x4 /vif;
run;
The output will show VIF for each variable. The higher the VIF, the more Multicollinearity and it is totally up to you to decide how big is big. If you have more variables to reduce, you may throw the ones with VIF > 2. On the other hand, if you don't have many variables, you may keep variables with VIF < 3.
@ Guoping Zeng : what is the title of your paper about multicollinearity?
thanks in adavanve
The paper below may give more insights on how to deal with multicollinearity:
Article Collinearity: A review of methods to deal with it and a simu...
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS