Collapses survey data to get direct estimates (i.e. non-pooled sample porportions)
direct_ests(.formula, .data, area_var, weight_var = NULL, shape = "wide")MRP formula. Only thing that will be used is the outcome variable (a binary variable)
Survey data to be collapsed
Character for the variable(s) that corresponds to the area to aggregate to.
Character for the variable that corresponds to weights.
Whether to return the output in "wide" (with one row per states)
or "long" (one row per state x estimate). Defaults to "wide".
A wide dataframe where each row is a area, p_raw indicates the
raw average, p_wt indicates the weighted average (if weight_var is provided),
and n_raw is the raw sample size. We also provide standard errors in the form se_se_raw follows the standard standard error for proportion sqrt(p*(1-p)/n).
The se_wt implements the weighted standard error, where the sample size
is replaced with the effective sample size, sum(wt)^2 / sum(wt^2).
direct_ests(response ~ (1|cd), cces_GA,
area_var = "cd",
weight_var = "weight_post")
#> # A tibble: 14 × 7
#> cd p_raw p_wt n_raw se_raw n_wt se_wt
#> <chr> <dbl> <dbl> <int> <dbl> <dbl> <dbl>
#> 1 GA-01 0.506 0.463 85 0.0542 43.6 0.0755
#> 2 GA-02 0.537 0.635 67 0.0609 30.5 0.0872
#> 3 GA-03 0.359 0.319 78 0.0543 60.2 0.0601
#> 4 GA-04 0.717 0.678 127 0.0400 69.8 0.0559
#> 5 GA-05 0.730 0.796 111 0.0422 30.7 0.0727
#> 6 GA-06 0.559 0.560 102 0.0492 78.6 0.0560
#> 7 GA-07 0.46 0.418 100 0.0498 49.8 0.0699
#> 8 GA-08 0.295 0.407 61 0.0584 33.3 0.0851
#> 9 GA-09 0.265 0.207 102 0.0437 72.1 0.0477
#> 10 GA-10 0.406 0.461 96 0.0501 60.7 0.0640
#> 11 GA-11 0.425 0.401 127 0.0439 68.7 0.0591
#> 12 GA-12 0.368 0.391 76 0.0553 47.5 0.0708
#> 13 GA-13 0.545 0.495 99 0.0500 52.2 0.0692
#> 14 GA-14 0.2 0.218 75 0.0462 31.5 0.0736