Cascade Survey Objects with Full Contingencies — bhhi

Calculates full contingent statistics in the manner of srvyr::cascade().

Usage

bhhi_cascade(.data, ..., .fill = "Overall")

Arguments

.data: tbl A tbl_svy object
...: Name-value pairs of summary functions
.fill: Value to fill in for group summaries. Defaults to 'Overall'.

Details

When calculating a two-way contingency table (a.k.a. crosstab), srvyr::cascade() calculates proportions of the column variable for each level of the row variable, but calculates the proportion of the row variable for the overall sample. This makes two-way contingency tables difficult with srvyr::cascade().

bhhi_cascade() instead calculates the proportion of the column variable for the overall sample, which produces results that are compatible and comparable to the results for the levels of the row variable

Examples

data("nhanes", package = "survey")

survey_object <- nhanes |>
  dplyr::rename(gender = RIAGENDR) |>
  dplyr::mutate(
    gender = factor(gender, 1:2, c("Male", "Female")),
    race = factor(race, 1:4, c("Hispanic", "White", "Black", "Other"))
  ) |>
  srvyr::as_survey(weights = WTMEC2YR)

survey_object |>
  srvyr::group_by(gender, race) |>
  bhhi_cascade(srvyr::survey_prop(proportion = TRUE))
#> # A tibble: 12 × 4
#> # Groups:   gender [3]
#>    gender  race       coef   `_se`
#>    <fct>   <fct>     <dbl>   <dbl>
#>  1 Male    Hispanic 0.158  0.00507
#>  2 Male    White    0.662  0.00818
#>  3 Male    Black    0.111  0.00432
#>  4 Male    Other    0.0682 0.00520
#>  5 Female  Hispanic 0.143  0.00443
#>  6 Female  White    0.653  0.00805
#>  7 Female  Black    0.127  0.00471
#>  8 Female  Other    0.0769 0.00536
#>  9 Overall Hispanic 0.151  0.00336
#> 10 Overall White    0.657  0.00574
#> 11 Overall Black    0.119  0.00321
#> 12 Overall Other    0.0726 0.00374

survey_object |>
  srvyr::group_by(gender, race) |>
  srvyr::cascade(srvyr::survey_prop(proportion = TRUE), .fill = "Overall")
#> Warning: There was 1 warning in `dplyr::summarise()`.
#> ℹ In argument: `srvyr::survey_prop(proportion = TRUE)`.
#> Caused by warning:
#> ! glm.fit: algorithm did not converge
#> # A tibble: 11 × 4
#> # Groups:   gender [3]
#>    gender  race       coef    `_se`
#>    <fct>   <fct>     <dbl>    <dbl>
#>  1 Male    Hispanic 0.158  5.07e- 3
#>  2 Male    White    0.662  8.18e- 3
#>  3 Male    Black    0.111  4.32e- 3
#>  4 Male    Other    0.0682 5.20e- 3
#>  5 Male    Overall  0.488  6.82e- 3
#>  6 Female  Hispanic 0.143  4.43e- 3
#>  7 Female  White    0.653  8.05e- 3
#>  8 Female  Black    0.127  4.71e- 3
#>  9 Female  Other    0.0769 5.36e- 3
#> 10 Female  Overall  0.512  6.82e- 3
#> 11 Overall Overall  1.00   9.26e-19