makeWeights

makeWeights is a function for creating tables of weighting values. It takes 3 input values: a dataframe (or data.table), a column with sample counts, and a column with population counts. The dataset should contain one row for each unique group requiring a weight.

Sample data

For testing, we use the HairEyeColor dataset that comes stock with R. It’s a table containing counts (Freq) for each combination of three categorical variables: Eye, Hair, and Sex. It looks something like this:

HairEyeColor
#> , , Sex = Male
#> 
#>        Eye
#> Hair    Brown Blue Hazel Green
#>   Black    32   11    10     3
#>   Brown    53   50    25    15
#>   Red      10   10     7     7
#>   Blond     3   30     5     8
#> 
#> , , Sex = Female
#> 
#>        Eye
#> Hair    Brown Blue Hazel Green
#>   Black    36    9     5     2
#>   Brown    66   34    29    14
#>   Red      16    7     7     7
#>   Blond     4   64     5     8

We tweak it by adding a random population variable. Next, we’ll apply the makeWeights function to it to look at what it adds onto our dataframe.

test_data <-  HairEyeColor %>%
  data.frame() %>%
  mutate(Population = round(runif(length(HairEyeColor),10000,20000),0))

test_dt <- data.table(test_data)

test_data %>% glimpse()
#> Rows: 32
#> Columns: 5
#> $ Hair       <fct> Black, Brown, Red, Blond, Black, Brown, Red, Blond, Black, …
#> $ Eye        <fct> Brown, Brown, Brown, Brown, Blue, Blue, Blue, Blue, Hazel, …
#> $ Sex        <fct> Male, Male, Male, Male, Male, Male, Male, Male, Male, Male,…
#> $ Freq       <dbl> 32, 53, 10, 3, 11, 50, 10, 30, 10, 25, 7, 5, 3, 15, 7, 8, 3…
#> $ Population <dbl> 11137, 16223, 16093, 16234, 18609, 16403, 10095, 12326, 166…

Sample table summary

The code chunk below applies makeWeights to the test data, summarizing the proportion and population weights for each record, along with some (optional) intermediate values for checking the calculations. Let’s look at the first few rows:

test_weights <- makeWeights(data = test_data, 
                            sampleVal = "Freq",
                            populationVal = "Population",
                            checkCols = TRUE)

knitr::kable(head(test_weights), digits = 2)

Hair	Eye	Sex	Freq	Population	sampleValWtd	populationWt	propWt	propCheck	populationCheck
Black	Brown	Male	32	11137	14.23	348.03	0.44	14.23	11137
Brown	Brown	Male	53	16223	20.73	306.09	0.39	20.73	16223
Red	Brown	Male	10	16093	20.57	1609.30	2.06	20.57	16093
Blond	Brown	Male	3	16234	20.75	5411.33	6.92	20.75	16234
Black	Blue	Male	11	18609	23.78	1691.73	2.16	23.78	18609
Brown	Blue	Male	50	16403	20.96	328.06	0.42	20.96	16403

The function calculates the proportion (propWt) which would transform the observed sample count (Freq) for a given row into a value (sampleVal_wtd) proportional to that row’s share of the overall population. It also produces the value (populationWt) that would transform the observed sample count into the overall population (Population) for a given row. Along the way it produces some intermediate values as well, for checking results.

Eventually, these checks will happen behind the scenes in more automated fashion, and would only pop up if the user requested them.