Analytical Covariance for SOPs, Average Times, and Comparisons

Materializes a selected covariance block from the low-rank analytical state retained by inferences(method = "delta"). Returns the conditional or unconditional covariance selected by the vcov argument to inferences(); it does not change that choice. Conditional covariance accounts for coefficient estimation with prediction covariates treated as given. Unconditional covariance also accounts for sampling the patients used in the average. Bootstrap and MVN results are not supported. With rows = NULL, the full dense covariance matrix is requested and its storage grows quadratically with the number of result cells. Use rows to extract smaller blocks within the analytical memory limit. Standard errors are computed without materializing the full covariance matrix.

Usage

# S3 method for class 'markov_sops'
vcov(object, rows = NULL, ...)

# S3 method for class 'markov_avg_sops'
vcov(object, rows = NULL, ...)

# S3 method for class 'markov_avg_comparisons'
vcov(object, rows = NULL, ...)

# S3 method for class 'markov_avg_time'
vcov(object, rows = NULL, ...)

Arguments

object

A markov_sops, markov_avg_sops, markov_avg_time, or markov_avg_comparisons object returned by inferences(method = "delta").

rows

Optional numeric or logical row selection, or analytical row-key values. NULL returns every result row.

...

Reserved for the stats::vcov() generic.

Value

A symmetric covariance matrix for the selected result cells.