R/plot_type_est.R

In sjPlot: Data Visualization for Statistics in Social Science

plot_type_est <- function(
  type,
  ci.lvl,
  se,
  tf,
  model,
  terms,
  group.terms,
  rm.terms,
  sort.est,
  title,
  axis.title,
  axis.labels,
  axis.lim,
  grid.breaks,
  show.intercept,
  show.values,
  show.p,
  value.offset,
  digits,
  geom.colors,
  geom.size,
  line.size,
  order.terms,
  vline.color,
  value.size,
  bpe,
  bpe.style,
  bpe.color,
  facets,
  show.zeroinf,
  p.threshold,
  p.val,
  vcov.fun,
  vcov.args,
  ci.style,
  p_adjust,
  std.response,
  ...
) {
  if (missing(facets)) {
    facets <- TRUE
  }

  # get tidy output of summary ----

  if (type == "std" || type == "std2") {
    std_method <- switch(type, "std" = "refit", "std2" = "2sd", "refit")
  } else {
    std_method <- FALSE
  }

  dat <- tidy_model(
    model = model,
    ci.lvl = ci.lvl,
    tf = tf,
    type = type,
    bpe = bpe,
    robust = list(vcov.fun = vcov.fun, vcov.args = vcov.args),
    facets = facets,
    show.zeroinf = show.zeroinf,
    p.val = p.val,
    standardize = std_method,
    bootstrap = FALSE,
    iterations = 1000,
    seed = NULL,
    p_adjust = p_adjust,
    std.response = std.response,
    ...
  )

  # fix brms coefficient names

  if (inherits(model, "brmsfit")) {
    dat$term <- gsub("^b_", "", dat$term)
  }

  # check if facet groups need to be replaced with title

  if (length(title) > 1) {
    tnames <- names(title)

    if (obj_has_name(dat, "facet") && !is.null(tnames)) {
      if (all(tnames %in% dat$facet)) {
        for (i in tnames) {
          dat$facet[which(dat$facet == i)] <- title[i]
        }
        title <- ""
      }
    }

    if (obj_has_name(dat, "response.level") && !is.null(tnames)) {
      if (all(tnames %in% dat$response.level)) {
        for (i in tnames) {
          dat$response.level[which(dat$response.level == i)] <- title[i]
        }
        title <- ""
      }
    }
  }

  # se needs to be logical from here on
  if (!is.null(se) && !is.logical(se)) {
    se <- TRUE
  }

  # for stan-models, we can define the style of the Bayesian point estimate,
  # which may be a line or a dot.

  if (missing(bpe.style) || is.null(bpe.style)) {
    bpe.style <- "line"
  }
  if (missing(value.size) || is.null(value.size)) {
    value.size <- 4
  }

  plot_model_estimates(
    model = model,
    dat = dat,
    tf = tf,
    se = se,
    terms = terms,
    group.terms = group.terms,
    rm.terms = rm.terms,
    sort.est = sort.est,
    title = title,
    axis.title = axis.title,
    axis.labels = axis.labels,
    axis.lim = axis.lim,
    grid.breaks = grid.breaks,
    show.intercept = show.intercept,
    show.values = show.values,
    show.p = show.p,
    value.offset = value.offset,
    digits = digits,
    geom.colors = geom.colors,
    geom.size = geom.size,
    line.size = line.size,
    bpe.style = bpe.style,
    bpe.color = bpe.color,
    term.order = order.terms,
    vline.color = vline.color,
    value.size = value.size,
    facets = facets,
    p.threshold = p.threshold,
    ci.style = ci.style,
    ...
  )
}