I am using the meologit command in Stata 13.1 to analyze an ordinal dependent variable with three categories (perceived job insecurity). The individual-level data is General Social Survey data from 1977-2012 (repeated cross-sectional data). I have 15,639 observations in this data set. My group-level data is at the state-level, and I have four state-level measures (unemployment, financialization, percent manufacturing, and unionization).
I am concerned about how I handle the state-level data with the cross-sections. I have tried a few different ways of structuring the command (different combinations of fixed and random effects):
The problem is that I have not found a good way to decide between these different options. Any advice would be greatly appreciated!
I am concerned about how I handle the state-level data with the cross-sections. I have tried a few different ways of structuring the command (different combinations of fixed and random effects):
Code:
meologit y x unemp finan pctmfg union i.year || state:, vce(cluster state) meologit y x unemp finan pctmfg union i.state || year:, vce(cluster state) meologit y x unemp finan pctmfg union || year: || state:, vce(cluster state)