Plot stacked Likert-style survey questions (ROME style)

Builds a horizontal 100% stacked bar chart for one or more survey questions, adding in-segment percent labels (hidden below a cutoff), a right-edge N label for each question, and a bottom legend that is left-aligned and wraps across rows when needed. Bar thickness and text size remain constant; the function computes and attaches recommended figure dimensions (in inches) that you can pass to ggplot2::ggsave().

Usage

plot_ROME_survey_questions(
  data,
  columns = NA,
  ordering = NA,
  percentage_cutoff = 10,
  label_size = 4.5,
  n_label_pad_text = -0.1,
  per_question_in = 0.5,
  base_height_in = 0.8,
  legend_allowance_in = 0.6,
  width_in = 8,
  y_label_wrap_width = 40,
  legend_wrap_chars = 80,
  right_margin_min_pt = 24
)

plot_ROME_survey_by_demographic(
  data,
  columns = NA,
  demographic = NULL,
  ordering = NA,
  percentage_cutoff = 10,
  label_size = 4.5,
  n_label_pad_text = -0.1,
  per_question_in = 0.5,
  base_height_in = 0.8,
  legend_allowance_in = 0.6,
  width_in = 8,
  y_label_wrap_width = 40,
  legend_wrap_chars = 80,
  right_margin_min_pt = 24
)

Arguments

data: A data frame where each selected column is a survey question and values are categorical responses (character or factor). NAs are treated as non-responses and excluded from the per-question N.
columns: Character or integer vector selecting the question columns; order matters (bars/facets appear top→bottom in this order). Default NA uses all columns in data.
ordering: Optional character vector giving the desired response order from most-positive to most-negative. Controls bar stacking and legend order (legend display is reversed so the first level appears on the left of the bar).
percentage_cutoff: Numeric in [0, 100]; in-bar percentage labels smaller than this are suppressed (0-dp, with a % sign).
label_size: Numeric text size (ggplot2 size units) for both the in-bar percentage labels and the right-edge N labels.
n_label_pad_text: Numeric passed to hjust for the N labels; negative values place the label slightly outside the 100\ values pull it inside the bar.
per_question_in, base_height_in, legend_allowance_in, width_in: Sizing controls (in inches) used to compute recommended figure dimensions stored in the returned object (see “Value”).
y_label_wrap_width: Integer; approximate wrap width (characters) for question (or demographic) labels after flipping.
legend_wrap_chars: Rough character budget per legend row; used to choose guide_legend(nrow = ...) so long legends wrap cleanly.
right_margin_min_pt: Minimum right plot margin (points) to reserve space for the right-edge N labels.
demographic: Optional character scalar naming the demographic column in data (e.g., "Gender"). When supplied, the plot facets by question (top→bottom in the order of columns) and draws one stacked bar per demographic level within each question. When NULL or NA, this function falls back to plot_ROME_survey_questions().

Value

A ggplot object. In addition to the usual plot methods, the object carries these attributes for saving at a consistent size:

attr(p, "width_in") — recommended width in inches;
attr(p, "height_in") — recommended height in inches;
attr(p, "rome_dims") — list with width, height, and units = "in".

Details

The percent axis spans 0–100 with major gridlines every 5\ 0/25/50/75/100. Axis ticks and axis lines are suppressed. The legend is placed at the bottom, left-aligned, and can wrap into multiple rows according to legend_wrap_chars. Per-question N (excluding NA) is printed at the right edge of each bar. Colours are drawn from get_OME_colours() for the number of observed response levels.

Note: this function uses theme settings such as legend.location = "plot" (honored by ggplot2 >= 3.5.0); on earlier versions that setting is ignored, but the plot will still render.

Important

The top-to-bottom order of questions follows the order of columns you pass in. Internally the function fixes factor levels (with coord_flip()) so the visual order matches the supplied columns.

Examples

# Minimal toy example (creates a small survey-like data frame)
if (requireNamespace("ggplot2", quietly = TRUE)) {
  set.seed(1)
  lvls <- c("Agree a lot","Agree a little","Neither",
            "Disagree a little","Disagree a lot")
  make_col <- function(n = 400) {
    sample(c(lvls, NA), n, replace = TRUE,
           prob = c(0.34, 0.29, 0.20, 0.12, 0.05, 0.00))
  }
  df <- data.frame(
    `Is Jake lovely?`  = make_col(),
    `Is Jake wonderful?`  = make_col(),
    `Is Jake just the best?`  = make_col(),
    `Is Jake the person who most inspires you?` = make_col()
  )

  # The order of 'columns' controls the top->bottom order in the chart:
  cols <- names(df)

  p <- plot_ROME_survey_questions(
    df,
    columns = cols,
    ordering = lvls,
    percentage_cutoff = 10,
    y_label_wrap_width = 40
  )
  print(p)

  # Save with the recommended size:
  ggplot2::ggsave(
    "rome_demo.png", p,
    width  = attr(p, "width_in"),
    height = attr(p, "height_in"),
    units  = "in", dpi = 300
  )
}


if (requireNamespace("ggplot2", quietly = TRUE)) {
  set.seed(1)
  lvls <- c("Agree a lot","Agree a little","Neither",
            "Disagree a little","Disagree a lot")
  make_col <- function(n = 400) {
    sample(c(lvls, NA), n, replace = TRUE,
           prob = c(0.34, 0.29, 0.20, 0.12, 0.05, 0.00))
  }

  df <- data.frame(
    `Is Jake lovely?`  = make_col(),
    `Is Jake wonderful?`  = make_col(),
    `Is Jake just the best?`  = make_col(),
    `Is Jake the person who most inspires you?` = make_col(),
    Gender = sample(c("Male","Female"), 400, replace = TRUE, prob = c(0.47, 0.53))
  )

  cols <- names(df)[names(df) != "Gender"]

  p_dem <- plot_ROME_survey_by_demographic(
    data = df,
    columns = cols,
    demographic = "Gender",
    ordering = lvls,
    percentage_cutoff = 10,
    y_label_wrap_width = 30
  )
  print(p_dem)

  ggplot2::ggsave(
    "rome_questions_by_gender.png", p_dem,
    width  = attr(p_dem, "width_in"),
    height = attr(p_dem, "height_in"),
    units  = "in", dpi = 300
  )
}