R/anova_tables.R

In dr-JT/resultsoutput: Get full output from statistical models in nice looking tables

#' ANOVA Tables
#'
#' Get ANOVA F-Table, contrasts, and pairwise comparisons
#'
#' @param x an lmer model object
#' @param eta_squared logical. Include partial-eta sqaured effect size?
#'     Default: TRUE
#' @param omega_squared logical. Include omega sqaured effect size?
#'     Default: TRUE
#' @param epsilon_squared logical. Include epsilon sqaured effect size?
#'     Default: FALSE
#' @param effects "fixed" or "all". default is "fixed" to reduce computation
#'     time
#' @param contrast The factor(s) at which to compare levels at
#' @param at Additional interacting factor(s) to compare the effect of contrast
#'      at
#' @param standardized Logical, indicating whether or not to print standardized
#'      estimates. Standardized estimates are based on "refit" of the model
#'      on standardized data but it will not standardize categorical predictors.
#'      Defualt is TRUE.
#' @param unstandardized Logical, indicating whether or not to print
#'      unstandardized estimates. Default is TRUE.
#' @param ci Confidence Interval (CI) level. Default to 0.95
#' @param ci_method Documention based on ?parameters::parameters.
#'     Method for computing degrees of freedom for confidence
#'     intervals (CI) and the related p-values. Allowed are following options
#'     (which vary depending on the model class): "residual", "normal",
#'     "likelihood", "satterthwaite", "kenward", "wald", "profile", "boot",
#'     "uniroot", "ml1", "betwithin", "hdi", "quantile", "ci", "eti", "si",
#'     "bci", or "bcai". See section Confidence intervals and approximation of
#'     degrees of freedom in model_parameters() for further details.
#'     When ci_method=NULL, in most cases "wald" is used then.
#' @param p_adjust The p-values adjustment method for frequentist multiple
#'     comparisons. Can be one of "holm", "tukey", "hochberg",
#'     "hommel", "bonferroni", "BH", "BY", "fdr" or "none" (default). See the
#'     p-value adjustment section in the emmeans::test documentation.
#' @param bootstrap Documention based on ?parameters::parameters.
#'     Should estimates be based on bootstrapped model? If TRUE, then arguments
#'     of Bayesian regressions apply (see also bootstrap_parameters()).
#' @param iterations Documention based on ?parameters::parameters.
#'     The number of bootstrap replicates. This only apply in the case of
#'     bootstrapped frequentist models.
#' @param pbkrtest.limit Optional parameter that can be set to help calculate
#'     dfs.
#'     If you need to use this a warning message will appear in the console
#'     telling you what to set this at.
#' @param lmerTest.limit Optional parameter that can be set to help calculate
#'     dfs.
#'     If you need to use this a warning message will appear in the console
#'     telling you what to set this at.
#' @param digits How many decimal places to round to? Default is 3.
#' @param id_col The column containing subject ids. Default is "Subject"
#' @param print Create a knitr table for displaying as html table?
#'     (default = TRUE)
#' @export
#'

anova_tables <- function(x,
                         eta_squared = TRUE,
                         omega_squared = TRUE,
                         epsilon_squared = FALSE,
                         effects = "fixed", contrast = NULL, at = NULL,
                         standardized = TRUE,
                         unstandardized = TRUE,
                         ci = 0.95, ci_method = NULL,
                         p_adjust = "none",
                         bootstrap = FALSE, iterations = NULL,
                         pbkrtest.limit = NULL,
                         lmerTest.limit = NULL,
                         digits = 3, id_col = "Subject",
                         print = TRUE) {

  table_modelsig <- anova_modelsig(x, eta_squared = eta_squared,
                                   omega_squared = omega_squared,
                                   epsilon_squared = epsilon_squared,
                                   digits = digits, id_col = id_col)
  tables <- gt::gt_group(table_modelsig)

  model_type <- insight::model_name(x)
  if (stringr::str_detect(model_type, "lmer")) {
    table_contrasts <- regression_coeff(x, effects = effects,
                                        standardized = standardized,
                                        unstandardized = unstandardized,
                                        ci = ci, ci_method = ci_method,
                                        bootstrap = bootstrap,
                                        iterations = iterations,
                                        digits = digits)
    tables <- gt::grp_add(tables, table_contrasts)
  }

  for (contr in contrast) {
    table_comparisons <- anova_comparisons(x, contrast = contr,
                                           p_adjust = p_adjust,
                                           digits = digits,
                                           pbkrtest.limit = pbkrtest.limit,
                                           lmerTest.limit = lmerTest.limit)
    tables <- gt::grp_add(tables, table_comparisons)
  }

  if (!is.null(at)) {
    for (contr in contrast) {
      interaction <- at[which(at != contr)]
      for (int in interaction) {
        table_comparisons <- anova_comparisons(x,
                                               contrast = contr,
                                               at = int,
                                               p_adjust = p_adjust,
                                               digit = digits,
                                               pbkrtest.limit =
                                                 pbkrtest.limit,
                                               lmerTest.limit =
                                                 lmerTest.limit)
        tables <- gt::grp_add(tables, table_comparisons)
      }
    }
  }
  return(tables)
}