V3 pct time canada

[rafdoodle]

Hello Doug and Caitlin,

Here is a new v3 PR for percent time in Canada, following up form PR #149.

In this PR, I added the following new variables:

• SDCGCB for 2001-2018 master (_m) data and 2009-2012 single-year shared (_s) data (replacing SDCGCBG_A)
• SDCDRES for 2001-2018 master (_m) data and 2009-2012 single-year shared (_s) data (replacing SDCGRES_A)

I also created a new function in R/percent-time-canada.R called pct_time_fun_A, which takes in the two above variables and DHH_AGE to calculate % time in Canada in the Master data.

• The key difference is that the new _A function simply does residency/age * 100 with the Master's continuous SDCDRES, unlike residency midpoint/age * 100 when using the PUMF's categorical SDCGRES.
• Tests were created for this new function in tests/testthat/test-immigarion.R.
• DHH_AGE's databaseStart and variableStart had to be updated to only _m and _s data in the worksheets for rec_with_table() to work which it did.

Please let me know what you think.

Thanks,
Rafidul

[DougManuel]

Code review

Reviewed 7 variables (DHH_AGE, SDCGCB, SDCGCBG, SDCDRES, SDCGRES, pct_time_der, pct_time_der_A) for PUMF and Master across 2001-2018.

L6 integration test: cross-cycle prevalence (PUMF)

rec_with_table() ran successfully for all 9 PUMF cycles. No step changes at era boundaries.

Cycle	pct_time_der valid %
cchs2001_p	100%
cchs2003_p	96.5%
cchs2005_p	93.5%
cchs2007_2008_p	100%
cchs2009_2010_p	98%
cchs2011_2012_p	95%
cchs2013_2014_p	96.5%
cchs2015_2016_p	95.5%
cchs2017_2018_p	98%

Master (_m) mappings validated by worksheet checks only — no PUMF data available for L6 testing of pct_time_der_A.

Issues found

6 issues (4 PR-introduced, 2 pre-existing/informational):

1. DHH_AGE databaseStart typo (P0): cchs2007_2008m missing underscore — should be cchs2007_2008_m. Affects variables.csv and all variable_details.csv rows.

https://github.com/Big-Life-Lab/cchsflow/blob/5b8954748947ddfd33269a4b6c48c6821c894d1d/inst/extdata/variables.csv

2. pct_time_der_A databaseStart double comma (P1): Contains , , (empty entry) between cchs2017_2018_m and cchs2009_s.

https://github.com/Big-Life-Lab/cchsflow/blob/5b8954748947ddfd33269a4b6c48c6821c894d1d/inst/extdata/variables.csv

3. DHH_AGE missing explicit 2015+ master mappings (P1): [DHH_AGE] default applies to cchs2015_2016_m and cchs2017_2018_m without explicit db::VAR mappings. Other SDC variables in this PR correctly have explicit 2015+ mappings (SDCDVCB, SDCDVRES, etc.). Needs DDI verification — DHH_AGE may not have been renamed, in which case the default is correct.

4. SDCGCB dummyVariable colon notation (P2): Uses _NA::a/_NA::b instead of _NAa/_NAb. Affects 6 rows in variable_details.csv. Pre-existing pattern in SDCGCBG and SDCGRES (on v3 branch) — fixing SDCGCB keeps new variables consistent with the updated naming convention.

5. Missing unit tests for born-in-Canada case (P1): Neither pct_time_fun() nor pct_time_fun_A() has a test for SDCGCBG=1 / SDCGCB=1 (returns 100). Also missing: pct_time_fun_cat(100) boundary test and pct_time_fun_cat(tagged_na("a")).

https://github.com/Big-Life-Lab/cchsflow/blob/5b8954748947ddfd33269a4b6c48c6821c894d1d/tests/testthat/test-immigration.R

6. Missing man/pct_time_fun_A.Rd (P2): Roxygen docs exist in source but .Rd file not generated. Run devtools::document().

cchsflow/R/percent-time-canada.R

Lines 78 to 118 in 5b89547

	#' @title Percent time in Canada (Master files)
	#'
	#' @description This function creates a derived variable (pct_time_der_A) that
	#' provides an estimated percentage of the time a person's life was spent in
	#' Canada. This variant is for Master file data where SDCDRES is continuous
	#' (actual years) rather than categorical.
	#'
	#' @param DHH_AGE continuous age variable.
	#'
	#' @param SDCGCB whether or not someone was born in Canada (1 - born in Canada,
	#' 2 - born outside Canada)
	#'
	#' @param SDCDRES continuous variable for how long someone has lived in Canada
	#' (in years).
	#'
	#' @return Numeric value between 0 and 100 that represents
	#' percentage of a respondent's time in Canada
	#'
	#' @examples
	#' # Using pct_time_fun_A() to generate a value for percent time spent in Canada
	#' # with user inputted values. Let's say you are 27 years old who was born
	#' # outside of Canada and have been living in Canada for 10 years.
	#' # Your estimated percent time spent in Canada can be calculated as follows:
	#'
	#' library(cchsflow)
	#' pct_time <- pct_time_fun_A(DHH_AGE = 27, SDCGCB = 2, SDCDRES = 10)
	#'
	#' print(pct_time)
	#'
	#' @export
	pct_time_fun_A <-
	function(DHH_AGE, SDCGCB, SDCDRES) {
	if (is_equal(SDCGCB, 1)) {
	return(100)
	}
	DHH_AGE <- if_else2(DHH_AGE > 0, DHH_AGE,
	return(tagged_na("b")))
	SDCDRES <- if_else2(SDCDRES >= 0, SDCDRES, return(tagged_na("b")))
	if_else2(SDCGCB == 2, (SDCDRES / DHH_AGE * 100), tagged_na("b"))
	}

Checked: era boundary defaults, databaseStart consistency, PUMF/Master naming, pre-2007 cycle letters, known error patterns, DV specifications, unit tests, and PUMF integration.

CEP: ceps/cep-008-immigration/

Generated with Claude Code

[rafdoodle]

DHH_AGE has maintained the same name since 2007-2008. Isn't it supposed to be _NA::a/_NA::b though? It's like that everywhere.

[DougManuel]

Code review — update after refactoring (2c4c9d1)

Reviewed 5 in-scope variables (SDCGRES_cont, pct_time_der, pct_time_der_cat10, SDCDRES, SDCGCB) for PUMF and Master across 2001-2018. Full review details in ceps/cep-008-immigration/PR-165-review-summary.md.

Summary of changes since last review

All 6 issues from the original review are resolved:

#	Issue	Resolution
1	DHH_AGE cchs2007_2008m typo (P0)	Fixed in 6260d65
2	pct_time_der_A double comma (P1)	Eliminated — pct_time_der_A merged into pct_time_der
3	DHH_AGE missing 2015+ mappings (P1→informational)	DDI verification confirms DHH_AGE unchanged through 2021; default is safe
4	SDCGCB dummyVariable colon notation (P2)	Pre-existing on v3 branch for SDCGCBG/SDCGRES; out of scope
5	Missing unit tests (P1)	28 tests now covering born-in-Canada, boundaries, vector inputs, output bounds
6	Missing pct_time_fun_A.Rd (P2)	Function removed — unified into calculate_pct_time()

Key refactoring in latest commit

• Unified DV function: pct_time_fun() and pct_time_fun_A() replaced by calculate_pct_time(age, born_in_canada, years_in_canada). pct_time_fun_cat() renamed to categorize_pct_time(). Follows v3 verb-first naming.
• Merged pct_time_der_A into pct_time_der: Separate PUMF and master row sets under one variable name. _s references purged.
• New SDCGRES_cont: PUMF categorical → continuous midpoint conversion moved from R code to worksheet. calculate_pct_time() receives continuous years for both PUMF and master.
• Output bounds validation: Values outside [0, 100] → tagged_na("b"). Documented in variable_details notes.
• case_when() replaces if_else2(): Vectorized logic, no early return().

L6 integration and tests

L6 PUMF integration (rec_with_table() across 9 cycles) ran successfully — no step changes at era boundaries, stable valid % (93.5-100%). See CEP docs for cross-cycle tables.

28/28 unit tests pass. Master mappings validated by worksheet checks only (no PUMF runtime data for _m).

Issues found in this review pass

No new issues. All L3-L5 checks pass (era boundaries, databaseStart consistency, naming conventions, error patterns, DV spec). Pre-existing issues (18 check_worksheet errors in variable_details.csv column ordering and row sorting) confirmed on v3 target branch — not introduced by this PR.

Checked: era boundary defaults, databaseStart consistency, PUMF/Master naming, pre-2007 cycle letters, known error patterns, DV specification, unit tests, PUMF integration, output bounds validation.

CEP: ceps/cep-008-immigration/