Skip to contents

Zimbabwe enumeration areas for demographic, health, and child-indicator surveys

Source: R/datasets.R
zwe_eas.Rd

A derived enumeration area (EA) frame for two-stage demographic, health, and child-indicator household surveys. Each row corresponds to one modeled WorldPop/GRID3 preEA polygon, and ea_id preserves the source identifier for spatial joins. Population and household counts are disaggregated from Zimbabwe's 2022 census ward totals. The frame covers 10 provinces and 91 districts.

Usage

zwe_eas

Format

A tibble with 107,250 rows and 12 columns:

ea_id

Integer. Unique preEA identifier from the WorldPop source

province

Factor. Province name (10 provinces)

district

Factor. District name within province (91 districts)

ward_pcode

Character. Ward P-code carried by the source preEA product (e.g. "ZW150104")

urban_rural

Factor. Modeled urban/rural classification based on building density and calibrated to GHS-DUC provincial shares

population

Integer. EA population, calibrated to 2022 Census ward totals

households

Integer. Number of households, calibrated to 2022 Census ward totals

buildings

Integer. Modeled building count from the WorldPop preEA product

women_15_49

Integer. Estimated women aged 15-49, from WorldPop age-sex grids scaled to census population

men_15_49

Integer. Estimated men aged 15-49, from WorldPop age-sex grids scaled to census population

children_under5

Integer. Estimated children under 5, from WorldPop age-sex grids scaled to census population

area_km2

Numeric. EA area in square kilometers

Source

Details

This dataset is designed for demonstrating:

  • Two-stage cluster sampling (EAs then households)

  • PPS sampling using household or population counts

  • Stratification by province and urban/rural

  • Partial execution (operational multi-stage sampling)

  • Creating household listings from selected EAs for second-stage sampling

The data structure follows typical two-stage cluster survey frames where EAs are nested within districts and provinces. To create a household listing for second-stage sampling after selecting EAs, expand each selected EA into individual household rows. The preEAs are modeled building-delimited areas rather than official census EAs, and the resulting counts should not be used as official small-area statistics.

# After stage 1 selection:
listing <- selected[rep(seq_len(nrow(selected)), selected$households), ]
listing$hh_id <- seq_len(nrow(listing))

See also

add_stage(), cluster_by(), execute()

Other datasets: bfa_eas, bfa_eas_cost, bfa_eas_variance, ken_enterprises, samplyr_datasets

Examples

# Explore the data
head(zwe_eas)
#> # A tibble: 6 × 12
#>   ea_id province district ward_pcode urban_rural population households buildings
#>   <int> <fct>    <fct>    <chr>      <fct>            <int>      <int>     <int>
#> 1   213 Bulawayo Bulawayo ZW102103   Urban              123         34        72
#> 2   214 Bulawayo Bulawayo ZW102103   Urban              105         29        62
#> 3   215 Bulawayo Bulawayo ZW102103   Urban              256         70       151
#> 4   216 Bulawayo Bulawayo ZW102103   Rural              193         53       113
#> 5   217 Bulawayo Bulawayo ZW102103   Urban              186         51       109
#> 6   218 Bulawayo Bulawayo ZW102103   Urban              212         58       125
#> # ℹ 4 more variables: women_15_49 <int>, men_15_49 <int>,
#> #   children_under5 <int>, area_km2 <dbl>
table(zwe_eas$province)
#> 
#>            Bulawayo              Harare          Manicaland Mashonaland Central 
#>                2164                6066               15287                9140 
#>    Mashonaland East    Mashonaland West            Masvingo  Matabeleland North 
#>               13632               14109               15312                8712 
#>  Matabeleland South            Midlands 
#>                8003               14825 
table(zwe_eas$urban_rural)
#> 
#> Rural Urban 
#> 83091 24159 

# Two-stage cluster sample: EAs then households
design <- sampling_design() |>
  add_stage(label = "EAs") |>
    stratify_by(province, urban_rural) |>
    cluster_by(ea_id) |>
    draw(n = 3, method = "pps_systematic", mos = households) |>
  add_stage(label = "Households") |>
    draw(n = 10)

selected <- execute(design, zwe_eas, stages = 1, seed = 123)

# A household listing of the selected EAs, made after fieldwork
library(dplyr)
listing <- selected |>
  as.data.frame() |>
  reframe(hh_id = seq_len(households), .by = ea_id)

# Continue the same design on the listing
smpl <- execute(selected, listing, seed = 1234)
smpl
#> # A tbl_sample: 600 × 12
#> # Sampling:     2 stages | 600 units
#> # Weights:      6357.88 [138.87, 21615.7]
#>    ea_id hh_id province urban_rural .weight .sample_id .stage .weight_2 .fpc_2
#>  * <int> <int> <fct>    <fct>         <dbl>      <int>  <int>     <dbl>  <dbl>
#>  1  1435    28 Bulawayo Urban         5818.          1      2       3.4     34
#>  2  1435    16 Bulawayo Urban         5818.          2      2       3.4     34
#>  3  1435    26 Bulawayo Urban         5818.          3      2       3.4     34
#>  4  1435    22 Bulawayo Urban         5818.          4      2       3.4     34
#>  5  1435     5 Bulawayo Urban         5818.          5      2       3.4     34
#>  6  1435    12 Bulawayo Urban         5818.          6      2       3.4     34
#>  7  1435    15 Bulawayo Urban         5818.          7      2       3.4     34
#>  8  1435     9 Bulawayo Urban         5818.          8      2       3.4     34
#>  9  1435    30 Bulawayo Urban         5818.          9      2       3.4     34
#> 10  1435     6 Bulawayo Urban         5818.         10      2       3.4     34
#> # ℹ 590 more rows
#> # ℹ 3 more variables: .weight_1 <dbl>, .fpc_1 <int>, .certainty_1 <lgl>