Skip to contents

Variance/Covariance for teeMod objects

Source: R/SandwichLayerVariance.R
var_estimators.Rd

Compute robust sandwich variance estimates with optional covariance adjustment

Usage

vcov_tee(x, type = NULL, cluster = NULL, ...)

.vcov_DB0(x, ...)

.vcov_DB(x, ...)

Arguments

x

a fitted teeMod model

type

a string indicating the desired bias correction for the residuals of x. Default makes no bias correction. See Details for supported types

cluster

a vector indicating the columns that define clusters. The default is the unit of assignment columns in the StudySpecification stored in x. These columns should appear in the dataframe used for fitting x as well as the dataframe passed to the covariance model fit in the case of prior covariance adjustment. See Details

...

arguments to be passed to the internal variance estimation function, such as cov_adj_rcorrect and loco_residuals. If x has a SandwichLayer object in its offset, The former specifies the bias correction to the residuals of the covariance model, and the latter indicates whether the offset should be replaced with predictions from leave-one-cluster-out fits of the covariance adjustment model. See Details

Value

A variance-covariance matrix with row and column entries for the estimated coefficients in x, the marginal mean outcome in the control condition, the marginal mean offset in the control condition (if an offset is provided), and if a moderator variable is specified in the formula for x, the mean interaction in the control condition of the outcome and offset with the moderator variable

Details

Variance estimates will be clustered on the basis of the columns provided to cluster (or obtained by the default behavior). As a result, providing "HCx" or "CRx" to type will produce the same variance estimate given that cluster remains the same.

With prior covariance adjustment, unless the data argument of the covariance model fit is the same as the data argument for fitting x and the StudySpecification of x has been created with a formula of the form trt_col ~ 1, the column(s) provided to cluster must appear in the dataframes in both data arguments, even if the clustering structure does not exist, per se, in the covariance adjustment sample. For instance, in a finely stratified randomized trial, one might desire standard errors clustered at the block level, but the covariance adjustment model may include auxiliary units that did not participate in the trial. In this case, in the data argument of the fitted covariance model, the column(s) passed to cluster should have the block ID's for rows overlapping with the data argument used for fitting x, and NA's for any auxiliary units. vcov_tee() will treat each row with an NA as its own cluster.

For ITT effect estimates without covariance adjustment, type corresponds to the variance estimate desired. Supported options include:

  • "MB0", "HC0", and "CR0" for model-based HC/CR0 standard errors

  • "MB1", "HC1", and "CR1" for model-based HC/CR1 standard errors (for "MB1" and "HC1", this is \(n/(n - 2)\), and for "CR1", this is \(g\cdot(n-1)/((g-1)\cdot(n-2))\), where \(g\) is the number of clusters in the sample used for fitting x)

  • "MB2", "HC2", and "CR2" for model-based HC/CR2 standard errors

  • "DB0" for design-based HC0 variance estimates

The type argument does not correspond to existing variance estimators in the literature in the case of prior covariance adjustment. It specifies the bias correction to the residuals of x, but the residuals of the covariance model are corrected separately based on the cov_adj_rcorrect argument. The cov_adj_rcorrect argument takes the same options as type except "DB0". When the covariance model includes rows in the treatment condition for fitting, the residuals of x are further corrected by having the values of offset replaced by predictions that use coefficient estimates that leave out rows in the same cluster (as defined by the cluster argument).

The design-based variance estimates can be calculated for teeMod models satisfying the following requirements:

  • The model uses rct_spec as StudySpecification

  • The model only estimates a main treatment effect

  • Inverse probability weighting is incorporated