| 1 |
#------------------------------------- |
| 2 |
# vcovCR with defaults |
| 3 |
#------------------------------------- |
| 4 |
|
| 5 |
#' Cluster-robust variance-covariance matrix for a glm object. |
| 6 |
#' |
| 7 |
#' \code{vcovCR} returns a sandwich estimate of the variance-covariance matrix |
| 8 |
#' of a set of regression coefficient estimates from an \code{\link{glm}} object. |
| 9 |
#' |
| 10 |
#' @param cluster Expression or vector indicating which observations belong to |
| 11 |
#' the same cluster. Required for \code{glm} objects. |
| 12 |
#' @param target Optional matrix or vector describing the working |
| 13 |
#' variance-covariance model used to calculate the \code{CR2} and \code{CR4} |
| 14 |
#' adjustment matrices. If a vector, the target matrix is assumed to be |
| 15 |
#' diagonal. If not specified, the target is taken to be the estimated variance function. |
| 16 |
#' @inheritParams vcovCR |
| 17 |
#' |
| 18 |
#' @return An object of class \code{c("vcovCR","clubSandwich")}, which consists |
| 19 |
#' of a matrix of the estimated variance of and covariances between the |
| 20 |
#' regression coefficient estimates. |
| 21 |
#' |
| 22 |
#' @seealso \code{\link{vcovCR}} |
| 23 |
#' |
| 24 |
#' @examples |
| 25 |
#' |
| 26 |
#' if (requireNamespace("geepack", quietly = TRUE)) { |
| 27 |
#' |
| 28 |
#' data(dietox, package = "geepack") |
| 29 |
#' dietox$Cu <- as.factor(dietox$Cu) |
| 30 |
#' weight_fit <- glm(Weight ~ Cu * poly(Time, 3), data=dietox, family = "quasipoisson") |
| 31 |
#' V_CR <- vcovCR(weight_fit, cluster = dietox$Pig, type = "CR2") |
| 32 |
#' coef_test(weight_fit, vcov = V_CR, test = "Satterthwaite") |
| 33 |
#' |
| 34 |
#' } |
| 35 |
#' |
| 36 |
#' @export |
| 37 |
|
| 38 |
vcovCR.glm <- function(obj, cluster, type, target = NULL, inverse_var = NULL, form = "sandwich", ...) { |
| 39 |
if (missing(cluster)) stop("You must specify a clustering variable.") |
| 40 |
if (is.null(inverse_var)) inverse_var <- is.null(target) |
| 41 |
vcov_CR(obj, cluster = cluster, type = type, |
| 42 |
target = target, inverse_var = inverse_var, form = form) |
| 43 |
} |
| 44 |
|
| 45 |
# coef() |
| 46 |
# nobs() |
| 47 |
|
| 48 |
#----------------------------------------------- |
| 49 |
# Model matrix |
| 50 |
#----------------------------------------------- |
| 51 |
|
| 52 |
#' @export |
| 53 |
|
| 54 |
model_matrix.glm <- function(obj) { |
| 55 |
X <- model.matrix(obj) |
| 56 |
eta <- obj$linear.predictors |
| 57 |
dmu_deta <- obj$family$mu.eta |
| 58 |
d <- dmu_deta(eta) |
| 59 |
d * X |
| 60 |
} |
| 61 |
|
| 62 |
#------------------------------------- |
| 63 |
# residuals |
| 64 |
#------------------------------------- |
| 65 |
|
| 66 |
#' @export |
| 67 |
|
| 68 |
residuals_CS.glm <- function(obj) { |
| 69 |
residuals(obj, type = "response") |
| 70 |
} |
| 71 |
|
| 72 |
#----------------------------------------------- |
| 73 |
# Get (model-based) working variance matrix |
| 74 |
#----------------------------------------------- |
| 75 |
|
| 76 |
#' @export |
| 77 |
|
| 78 |
targetVariance.glm <- function(obj, cluster) { |
| 79 |
mu <- fitted.values(obj) |
| 80 |
var_fun <- obj$family$variance |
| 81 |
v <- var_fun(mu) |
| 82 |
w <- weights(obj, type = "prior") |
| 83 |
matrix_list(v / w, cluster, "both") |
| 84 |
} |
| 85 |
|
| 86 |
#------------------------------------- |
| 87 |
# Get weighting matrix |
| 88 |
#------------------------------------- |
| 89 |
|
| 90 |
#' @export |
| 91 |
|
| 92 |
weightMatrix.glm <- function(obj, cluster) { |
| 93 |
mu <- fitted.values(obj) |
| 94 |
var_fun <- obj$family$variance |
| 95 |
v <- var_fun(mu) |
| 96 |
w <- weights(obj, type = "prior") |
| 97 |
matrix_list(w / v, cluster, "both") |
| 98 |
} |
| 99 |
|
| 100 |
#--------------------------------------- |
| 101 |
# Get bread matrix and scaling constant |
| 102 |
#--------------------------------------- |
| 103 |
|
| 104 |
# bread.glm() is in sandwich package |
| 105 |
|
| 106 |
#' @export |
| 107 |
|
| 108 |
v_scale.glm <- function(obj) { |
| 109 |
if (substr(obj$family$family, 1, 17) %in% c("poisson", "binomial", "Negative Binomial")) { |
| 110 |
dispersion <- 1 |
| 111 |
} else { |
| 112 |
wres <- as.vector(residuals(obj, "working")) * weights(obj, "working") |
| 113 |
dispersion <- sum(wres^2)/sum(weights(obj, "working")) |
| 114 |
} |
| 115 |
as.vector(sum(summary(obj)$df[1:2])) * dispersion |
| 116 |
} |