parse_lrt_table: parse_lrt_table

Description Usage Arguments Details

View source: R/lrt_contrast_set_io.R

Description

Each likelihood-ratio test is performed over a set of contrasts. This function extracts the names of the contrasts over which each LRT should be performed from a data.frame. The collection of contrasts should be defined as a ';'-separated string (or list of character-vectors) in the input data.frame.

Usage

1

Arguments

x

A data-frame containing (at least) columns "lrt_name", "level", "parent", "contrast_set". The 'contrast_set' column must be a vector of ';'-separated characters or a list of character-vectors.

Details

A tibble where the contrast sets are present in a list-column is returned.

The number of siblings of each LRT-node is also determined.

The 'level' should be 1 for the LRT-nodes comprised of the largest collection of contrasts and then increments for each proper subset of these nodes. level-1 nodes have no parent, so set parent = NA for these rows.

The tibble that is returned is ordered by increasing 'level', to try to ensure that parent nodes occur before children.

We allow a node to have at most one parent. For use in nested-LRT analysis, any node that has a missing (NA) parent must be at level-1.


russHyde/reeq documentation built on March 10, 2020, 5:36 a.m.