Skip to contents

Extracts coefficients and confidence intervals from linear models

Source: R/lm-tab.R
dh.lmTab.Rd

To conveniently view model results or make tables, it is useful to extract the coefficients in a useable format. This function extracts coefficients for ds.glm, ds.glmSLMA and ds.lmerSLMA objects.

Usage

dh.lmTab(
  model = NULL,
  type = NULL,
  coh_names = NULL,
  direction = NULL,
  ci_format = NULL,
  family = "gaussian",
  digits = 2,
  exponentiate = FALSE,
  extract_random = FALSE
)

Arguments

model

Object returned from either ds.glm, ds.glmSLMA or ds.lmerSLMA functions.

type

Character specifying type of object provided in model. Can be either "glm_ipd", "glm_slma" or "lmer_slma".

coh_names

Character vector of cohorts included in model. Note this must be in the same order as the cohorts were included in the model.

direction

Character specifying the output format. Can be either "wide" or "long". See "Value" for more details.

ci_format

format for the confidence intervals when direction is "wide". If "separate", upper and lower confidence intervals are displayed as columns. If "paste", confidence intervals are returned in the same column as the coefficient within brackets. Option "paste" is only available if ci_format is "wide"

family

Specifies the family used in the analysis where type is "glm_ipd" or "glm_slma". Options are "gaussian" or "binomial", with default "gaussian".

digits

Optionally, the number of decimal places to round output to. Default is 2.

exponentiate

Optionally, specify whether estimates from binomial models should be exponentiated, ie returned as odds ratios. This argument is ignored if type is "gaussian".

extract_random

Optionally, specify whether to return random effects in mixed effects models.

Value

A tibble. When direction is "wide" & ci_format is "paste", this contains five columns:

  • variable

  • est

  • lowci

  • uppci

  • pvalue

When direction is "long", a tibble with three columns is returned:

  • variable

  • coefficient (containing values "est", "lowci", "uppci", "pvalue")

  • value

When type is "lmer_slma", a list of 2 elements where the first element is as described as above, and the second element is a tibble of random effects with 5 columns: "cohort", "group", "var1", "var2", "stddev". "stddev" gives either the

#' Extracts the random effects from an lmer_slma model. Returns tibble containing, for each cohort, the standard deviation of the random terms and the standard deviation of the residual error. If two or more random terms are included, it also returns the correlation between the terms.

See also

Other descriptive functions: dh.anyData(), dh.classDiscrepancy(), dh.createTableOne(), dh.getStats(), dh.meanByGroup()