Add correspondence functions

[wfmackey]

From and to various versions of structures (eg ASCO to ANZSCO).

Could be population-weighted options for imperfect correspondence?

[cynthiahqy]

@wfmackey I might be able to help with this if someone can:
a) point me to a commonly needed transformation (e.g. from ASCO to ANZSCO?) -- ideally single step
b) point me to the relevant correspondence table (assuming it exists), or talk me through the sensible way to do the correspondence (e.g. population weights)
c) tell me how to get the relevant data

I'm writing some how-to vignettes for my package {conformr}, and have been looking for some open license data that needs concording for analysis to use as an example. I don't know much about ABS data, but as long as I could use the data & concordances in my vignettes, I'm happy to invest some time into building out these functions for {strayr}.

[wfmackey]

thanks @cynthiahqy ! A good place to start might be ASCO to ANZSCO (as you suggested).

Detailed correspondence tables between ANZSCO Revision 1 and ASCO Second Edition can be found at that link. Note that this is for anzsco2009.

You can of course them for the {conformr} package -- which looks like it'll be incredibly useful!

[cynthiahqy]

cheers @wfmackey, where can I find data reported in the ASCO structure -- an ABS dataset I presume?

[wfmackey]

sorry for the slow reply @cynthiahqy. It looks like getting ASCO in a usable format will be more difficult than I first thought. I'll keep searching and add it at some point.

For the mean time, how about ANZSCO-2021 to ISCO-08? The ISCO-08 file can be found here: https://www.ilo.org/ilostat-files/ISCO/newdocs-08-2021/ISCO-08/ISCO-08%20EN.csv
And the correspondence table can be found here: https://www.abs.gov.au/statistics/classifications/anzsco-australian-and-new-zealand-standard-classification-occupations/2021/anzsco%202021%20correspondence.xlsx

Will that work for you?

[cynthiahqy]

@wfmackey no rush at all.

Just to clarify, what kind of correspondence function do you have in mind?:

1. a per-code lookup -- e.g. take ONE ANZSCO6 digit, and correspond it to ALL matching ISCO-08 code, independent of data? -- i.e. something similar to concordance::concord()
2. OR data transformation -- i.e. take DATA encoded at the ANZSCO6 level and spit it out as ISIC-08 data?

[wfmackey]

I had number 1 in mind, where the user would provide a vector of original codes/names and gets corresponded codes or names in return.

But I am interested in how the second option would work!