I’ve created a package to help people put together data from the Panel Study of Income Dynamics (PSID). Given a simple JSON file describing the variables you want, PSID.jl will look up all the years that variable is available, identify and code missing values, generate consistent labels across time, check whether a variable is meant for the household head or household spouse, and create a panel of individuals linked across time with a consistent ID.
Instructions are available here: https://github.com/aaowens/PSID.jl
It’s not registered or well tested at this point, but it seems to work fine for the people I know who tried it.
I’ve added tests, CI, and package versioning. An update to JSON3 had broken PSID.jl, but that is now fixed.
There have been a few updates. The constructed data now includes the original variable codes as well, in case you want to look the variable up on the PSID’s website. The 1968 ID is now included, which is needed for sample selection like dropping the SEO sample.
I wrote a blog post using it to demonstrate what you can do with PSID data, looking at the relationship between income in your 30’s and income in your 50’s.
I registered this package, so you can just do
add PSID and then download the data files manually.