Summarize the replicate weights of a design
Arguments
- rep_design
A replicate design object, created with either the
surveyorsrvyrpackages.- type
Default is
"both". Usetype = "overall", for an overall summary of the replicate weights. Usetype = "specific"for a summary of each column of replicate weights, with each column of replicate weights summarized in a given row of the summary.
Usetype = "both"for a list containing both summaries, with the list containing the names"overall"and"both".- by
(Optional) A character vector with the names of variables used to group the summaries.
Value
If type = "both" (the default), the result is a list of data frames
with names "overall" and "specific". If type = "overall", the result is
a data frame providing an overall summary of the replicate weights.
The contents of the "overall" summary are the following:
"nrows": Number of rows for the weights
"ncols": Number of columns of replicate weights
"degf_svy_pkg": The degrees of freedom according to the survey package in R
"rank": The matrix rank as determined by a QR decomposition
"avg_wgt_sum": The average column sum
"sd_wgt_sums": The standard deviation of the column sums
"min_rep_wgt": The minimum value of any replicate weight
"max_rep_wgt": The maximum value of any replicate weight
If type = "specific", the result is a data frame providing a
summary of each column of replicate weights, with each column of replicate weights
described in a given row of the data frame.
The contents of the "specific" summary are the following:
"Rep_Column": The name of a given column of replicate weights. If columns are unnamed, the column number is used instead
"N": The number of entries
"N_NONZERO": The number of nonzero entries
"SUM": The sum of the weights
"MEAN": The average of the weights
"CV": The coefficient of variation of the weights (standard deviation divided by mean)
"MIN": The minimum weight
"MAX": The maximum weight
Examples
# Load example data
suppressPackageStartupMessages(library(survey))
data(api)
dclus1 <- svydesign(id=~dnum, weights=~pw, data=apiclus1, fpc=~fpc)
dclus1$variables$response_status <- sample(x = c("Respondent", "Nonrespondent",
"Ineligible", "Unknown eligibility"),
size = nrow(dclus1),
replace = TRUE)
rep_design <- as.svrepdesign(dclus1)
# Adjust weights for cases with unknown eligibility
ue_adjusted_design <- redistribute_weights(
design = rep_design,
reduce_if = response_status %in% c("Unknown eligibility"),
increase_if = !response_status %in% c("Unknown eligibility"),
by = c("stype")
)
# Summarize replicate weights
summarize_rep_weights(rep_design, type = "both")
#> $overall
#> nrows ncols degf_svy_pkg rank avg_wgt_sum sd_wgt_sums min_rep_wgt max_rep_wgt
#> 1 183 15 14 15 6194 403.1741 0 36.26464
#>
#> $specific
#> Rep_Column N N_NONZERO SUM MEAN CV MIN MAX
#> 1 1 183 172 6237.518 34.08480 0.25358407 0 36.26464
#> 2 2 183 179 6491.370 35.47197 0.14989713 0 36.26464
#> 3 3 183 181 6563.900 35.86830 0.10540606 0 36.26464
#> 4 4 183 170 6164.989 33.68846 0.27729183 0 36.26464
#> 5 5 183 181 6563.900 35.86830 0.10540606 0 36.26464
#> 6 6 183 179 6491.370 35.47197 0.14989713 0 36.26464
#> 7 7 183 179 6491.370 35.47197 0.14989713 0 36.26464
#> 8 8 183 167 6056.195 33.09396 0.31037848 0 36.26464
#> 9 9 183 174 6310.047 34.48113 0.22805336 0 36.26464
#> 10 10 183 149 5403.431 29.52695 0.47900073 0 36.26464
#> 11 11 183 162 5874.872 32.10312 0.36102892 0 36.26464
#> 12 12 183 146 5294.637 28.93244 0.50479412 0 36.26464
#> 13 13 183 170 6164.989 33.68846 0.27729183 0 36.26464
#> 14 14 183 182 6600.164 36.06647 0.07432829 0 36.26464
#> 15 15 183 171 6201.253 33.88663 0.26563324 0 36.26464
#>
# Summarize replicate weights by grouping variables
summarize_rep_weights(ue_adjusted_design, type = 'overall',
by = c("response_status"))
#> response_status nrows ncols degf_svy_pkg rank avg_wgt_sum sd_wgt_sums
#> 1 Ineligible 39 15 13 14 1896.620 164.4527
#> 2 Nonrespondent 47 15 14 15 2296.912 133.9403
#> 3 Respondent 41 15 13 14 2000.468 130.3750
#> 4 Unknown eligibility 56 15 -1 0 0.000 0.0000
#> min_rep_wgt max_rep_wgt
#> 1 0 56.98729
#> 2 0 56.98729
#> 3 0 56.98729
#> 4 0 0.00000
summarize_rep_weights(ue_adjusted_design, type = 'overall',
by = c("stype", "response_status"))
#> stype response_status nrows ncols degf_svy_pkg rank avg_wgt_sum
#> 1 E Ineligible 29 15 7 8 1413.77685
#> 2 H Ineligible 6 15 2 3 283.80822
#> 3 M Ineligible 4 15 3 4 199.03463
#> 4 E Nonrespondent 36 15 12 13 1753.97013
#> 5 H Nonrespondent 2 15 1 2 95.02487
#> 6 M Nonrespondent 9 15 5 6 447.91713
#> 7 E Respondent 35 15 10 11 1706.22048
#> 8 H Respondent 2 15 1 2 95.02487
#> 9 M Respondent 4 15 2 3 199.22315
#> 10 E Unknown eligibility 44 15 -1 0 0.00000
#> 11 H Unknown eligibility 4 15 -1 0 0.00000
#> 12 M Unknown eligibility 8 15 -1 0 0.00000
#> sd_wgt_sums min_rep_wgt max_rep_wgt
#> 1 149.65790 0 53.40068
#> 2 40.20252 0 56.98729
#> 3 23.57306 0 56.98729
#> 4 121.79339 0 53.40068
#> 5 18.18279 0 56.98729
#> 6 46.50443 0 56.98729
#> 7 135.21908 0 53.40068
#> 8 18.18279 0 56.98729
#> 9 32.77474 0 56.98729
#> 10 0.00000 0 0.00000
#> 11 0.00000 0 0.00000
#> 12 0.00000 0 0.00000
# Compare replicate weights
rep_wt_summaries <- lapply(list('original' = rep_design,
'adjusted' = ue_adjusted_design),
summarize_rep_weights,
type = "overall")
print(rep_wt_summaries)
#> $original
#> nrows ncols degf_svy_pkg rank avg_wgt_sum sd_wgt_sums min_rep_wgt max_rep_wgt
#> 1 183 15 14 15 6194 403.1741 0 36.26464
#>
#> $adjusted
#> nrows ncols degf_svy_pkg rank avg_wgt_sum sd_wgt_sums min_rep_wgt max_rep_wgt
#> 1 183 15 14 15 6194 403.1741 0 56.98729
#>