Why does multicollinearity happen when I convert the variables into categorical variables?

43 Views Asked by At

I'm trying to analyze the data with conditional logit model.

When I converted the integer variables into categorical variables, some of the results are omitted because of the collinearity.

Why does this happen and how can I solve this problem? I tried to find the correlation between the variables, but it seems all variables have the correlation. I got lost from here :/

The below coding are just the examples:

data2$Introduction <- as.factor(data2$Introduction)
clogout2 <- clogit(RES~Introduction+Waterside+Play+Wetland+WTP+strata(STR), data=data2)

sample result:

Waterside3           NA        NA  0.000000      NA       NA

I tried to find the vif: but all variables are saying: there are aliased coefficients in the model.

Also, tried to see the chi-square, it also showed the significant p-value among all variables.

I also tried to convert the variables into categorical without dummy variables("0", "1"). But couldn't find the correct coding so far...

0

There are 0 best solutions below