Splitting string variable 'Sector' into multiple variables.

[Dillan Patel]

Hi, I have a string variable called 'sector' consisting of 9 different industry sectors. I want to split this variable into its 9 sub-sectors so I can then interact with each sub-sector with union to see whether union/non-union wages are higher.

Does anybody know how I can split this string variable into 9 different variables for the different sectors?

[Dillan Patel]

Below is the STATA output

[Nick Cox]

separate can do this, but in practice it is perhaps as or more likely that you would be better off using your sector variable to separate other variables with numeric values.

[Dillan Patel]

Thank you for the response! For the second part, does that entail creating dummy variables for each sector (i.e., generating a variable such that Managers, Directors and Senior Office ==1)?

[Nick Cox]

separate is the tool of choice here for separating variables, which was your first question.

For your regression problem, you don't need to separate or use indicator variables (dummy variables in your terminology) at all.

encode sector, gen(num_sector) 

regress wages i.union i.num_sector

with added interaction terms. See documentation on factor variables for that,

Wages are often better analysed on logarithmic scale.