Code review: 01-demographic.R

[martindanka]

Fixes

• Ethnicity: Changed the mapping for code -94 from "Script error/information lost" to "Don't know/insufficient information" (now coded as -8) to reflect the documentation.
• Sexual orientation:
  • For S6 and S7 (sori19, sori20), codes -100 and -97 (declined sexual experience questions / refused self-completion) now map to -9 (“Refusal”) rather than -3.
  • Code -91 is now consistently treated as -1 (“Item not applicable”) instead of -8 (“Don’t know / insufficient information”).
  • For S8 (sori25), code -1 (“Not applicable”) is now kept as -1 (“Item not applicable”) instead of being merged into -8 ("Don't know/insufficient information").
  • For S9 (sori32), code 5 (“Prefer not to say”) is now mapped to -7 ("Prefer not to say"), which is consistent with documentation. Code -1 is now kept as “Item not applicable” rather than merged into -8 ("Don't know/insufficient information").
• Partnership status: Re-written using look-up tables to make the mapping very explicit. This was needed because the source variables are inconsistent in values/labels, so some substantive and missing categories were getting remapped incorrectly. I have also added cross-checks for the derived variables.

Improvements

• All derived variables now stored as haven::labelled rather than factors, so that values/labels match the documentation.
• Removed redundant is.na() checks and simplified some case_when() logic.