Hello everyone,
I am currently trying to analyze the survey data I collected for my master's thesis in Economics. I conducted a discrete choice experiment among University students who faced 4 hypothetical job scenarios, each consisting of three job offer alternatives. The participants had to provide a personal ranking between the three job offers, where 1 indicated the most preferred job offer and 3 the least. The job offers differed in 5 job attributes: Salary, Working Hours, Part Time Possibility, Working From Home Option, and CSR.
Salary has 9 attribute levels, Working Hours 5, CSR has 6, and Part Time and working from home are binary variables.
I want to estimate the effect of the job attributes on the choice of the students. To do this I ran this mixlogit command:
xi: mixlogit choice i.Salary, group(IDscreencset_num) id(ParticipantID) rand(i.PT i.CSR i.HomeOffice i.WorkingHours ) nrep(300) robust corr difficult
However, I receive the following error message: Some variables are collinear - check your model specification
The mixlogit command works if I don't include the i. But I am particularly interested in the specific effect of the attribute levels.
So far I tried multiple things to overcome this issue. I tried to manually create the binary variables for the attribute levels and exclude the basecategory. I also tried different baselevels. But nothing worked so far and I am losing my mind!!
If anyone could help me out here I would really appreciate it!!!!
I am currently trying to analyze the survey data I collected for my master's thesis in Economics. I conducted a discrete choice experiment among University students who faced 4 hypothetical job scenarios, each consisting of three job offer alternatives. The participants had to provide a personal ranking between the three job offers, where 1 indicated the most preferred job offer and 3 the least. The job offers differed in 5 job attributes: Salary, Working Hours, Part Time Possibility, Working From Home Option, and CSR.
Salary has 9 attribute levels, Working Hours 5, CSR has 6, and Part Time and working from home are binary variables.
I want to estimate the effect of the job attributes on the choice of the students. To do this I ran this mixlogit command:
xi: mixlogit choice i.Salary, group(IDscreencset_num) id(ParticipantID) rand(i.PT i.CSR i.HomeOffice i.WorkingHours ) nrep(300) robust corr difficult
However, I receive the following error message: Some variables are collinear - check your model specification
The mixlogit command works if I don't include the i. But I am particularly interested in the specific effect of the attribute levels.
So far I tried multiple things to overcome this issue. I tried to manually create the binary variables for the attribute levels and exclude the basecategory. I also tried different baselevels. But nothing worked so far and I am losing my mind!!
If anyone could help me out here I would really appreciate it!!!!
