Various age fields include end points when names and documentation indicate otherwise

[MaxGhenis]

Documentation and variable name structure for various age fields indicate they should not include the end point. I see four contradictions in the code on this front:

• nu05 includes 5-year-olds based on the finalprep.py code.
• nu13 includes 13-year-olds based on the finalprep.py code
• elderly_dependent is described as "over 65 years old" in the documentation, but includes 65-year-olds based on the finalprep.py code. As an aside, it would be helpful if the user documentation specified the age threshold for this variable.
• n24 has a related issue: only children under age 17 are eligible for the CTC, but finalprep.py includes 17-year-olds in n24.

n24 aside, I'm not sure if these are used elsewhere, and how that affects whether the variable names or the logic should be changed.

Relevant links:

• Does nu18 or n1821 include 18-year-olds? Tax-Calculator#1669.
• http://open-source-economics.github.io/Tax-Calculator/
• https://github.com/open-source-economics/taxdata/blob/master/cps_data/cps_file_doc.md

[MaxGhenis]

A process tweak that could help avoid this is to use exclusively either </> or <=/>= in code. finalprep.py includes endpoints for these variables, but just below them it doesn't with the f2441 and EIC variables (both defined correctly).

[martinholmer]

@andersonfrailey,
Where do we stand on correcting these bugs in the age variables in the cps.csv file?
Are the same kind of errors also present in the puf.csv file?

@MattHJensen @MaxGhenis

[andersonfrailey]

I'll put together a PR to address these issues today. In the PUF nu13 and nu05 include dependents age 13 and 5, so that will need to be updated.

[andersonfrailey]

@MaxGhenis, where in the documentation did you see elderly_dependent described as "over 65 years old"?

[MaxGhenis]

We then count the number of dependents under 5 and 13 or over 65 years old.

From https://github.com/open-source-economics/taxdata/blob/master/cps_data/cps_file_doc.md

[martinholmer]

@MaxGhenis, Thanks for pointing out these bugs in the generation of the CPS age variables.
These problems have been fixed in pull request #172.