randomizr (version 2.0.1)

strata_and_cluster_rs_probabilities: Inclusion probabilities: Stratified and Clustered Random Sampling

Description

Returns each unit's probability of being sampled when clusters are drawn within strata. Probabilities vary across strata and are constant within a cluster.

Usage

strata_and_cluster_rs_probabilities(
  strata = NULL,
  clusters = NULL,
  prob = NULL,
  prob_unit = NULL,
  n = NULL,
  n_unit = NULL,
  strata_n = NULL,
  strata_prob = NULL,
  check_inputs = TRUE
)

Value

A numeric vector of length N giving each unit's probability of being included in the sample. Every unit in a cluster shares one probability.

Arguments

strata

A vector of length N indicating which stratum each unit belongs to. Every unit in a cluster must belong to the same stratum. (required)

clusters

A vector of length N indicating which cluster each unit belongs to. (required)

prob

Use for a design in which either floor(N_clusters_stratum*prob) or ceiling(N_clusters_stratum*prob) clusters are sampled within each stratum. Which of the two is used is itself random: the ceiling is drawn with probability equal to the fractional part of N_clusters_stratum*prob and the floor otherwise, which makes each cluster's probability of inclusion exactly prob. Must be a real number between 0 and 1 inclusive. (optional)

prob_unit

Must be of length N. tapply(prob_unit, strata, unique) will be passed to strata_prob, so it must be constant within each stratum. (optional)

n

Use for a design in which the scalar n gives the fixed number of clusters to sample in every stratum. This count does not vary across strata. (optional)

n_unit

Must be of length N. tapply(n_unit, strata, unique) will be passed to strata_n, so it must be constant within each stratum. (optional)

strata_n

Use for a design in which strata_n gives the number of clusters to sample within each stratum. Must be as long as the number of strata, in the same order as sort(unique(strata)). (optional)

strata_prob

Use for a design in which strata_prob gives the probability of being sampled within each stratum. Must be in the same order as sort(unique(strata)). Differs from prob in that the probability of being sampled can vary across strata. (optional)

check_inputs

Logical. Whether to verify before sampling that the arguments are internally consistent: that clusters nest within strata, that counts do not exceed the number of clusters in a stratum, that probabilities lie between 0 and 1, and so on. Defaults to TRUE. Set to FALSE to skip the checks when drawing many samples from arguments that have already been verified; declaring the design once with declare_rs() and drawing from it with draw_rs() does this for you. (optional)

Details

These are the quantities inverse-probability weights are built from: weight each sampled unit by the reciprocal of its inclusion probability, which obtain_inclusion_probabilities() extracts for you.

See Also

strata_and_cluster_rs()

Examples

# Twelve clusters, of sizes 1 through 12, nested in four strata of three
clusters <- rep(letters[1:12], times = 1:12)

strata <- rep(NA, length(clusters))
strata[clusters %in% letters[1:3]] <- "stratum_1"
strata[clusters %in% letters[4:6]] <- "stratum_2"
strata[clusters %in% letters[7:9]] <- "stratum_3"
strata[clusters %in% letters[10:12]] <- "stratum_4"

table(strata, clusters)

probs <- strata_and_cluster_rs_probabilities(strata = strata,
                                         clusters = clusters)

table(probs, strata)
table(probs, clusters)


probs <- strata_and_cluster_rs_probabilities(clusters = clusters,
                                         strata = strata,
                                         prob = 0.5)

table(probs, clusters)
table(probs, strata)

probs <- strata_and_cluster_rs_probabilities(clusters = clusters,
                                         strata = strata,
                                         strata_n = c(1, 2, 1, 2))

table(probs, clusters)
table(probs, strata)

probs <- strata_and_cluster_rs_probabilities(clusters = clusters,
                                         strata = strata,
                                         strata_prob = c(0.2, 0.4, 0.6, 0.8))

table(probs, clusters)
table(probs, strata)