Skip to contents

Plot a probability of outcome superiority curve

Source: R/methods.R
autoplot.posc.Rd

autoplot() returns a ggplot2::ggplot() object that can be modified further; plot() draws it.

Usage

# S3 method for class 'posc'
autoplot(
  object,
  full = FALSE,
  color = "#1F5A96",
  ci = TRUE,
  reference = FALSE,
  bins = FALSE,
  percent = FALSE,
  grid = TRUE,
  mark = NULL,
  mark_ci = FALSE,
  title = NULL,
  subtitle = NULL,
  ...
)

# S3 method for class 'posc'
plot(x, ...)

Arguments

object, x

A "posc" object.

full

If FALSE (default), plot differences \(\Delta \ge 0\), where the curve runs from 0.5 upwards (or downwards for a negative relationship). If TRUE, plot the full sigmoid over negative and positive differences. The two halves are mirror images, since \(P(-\Delta) = 1 - P(\Delta)\).

color

Colour of the curve and its confidence band.

ci

Draw the confidence band (if one is available)?

reference

For empirical curves from posc_emp(): also draw the bivariate-normal POSC with the same Pearson correlation, as a dashed line? Off by default.

bins

For empirical curves fitted with bins: draw the binned observed proportions as points, with whiskers showing simultaneous confidence intervals (see posc_bins())? Off by default.

percent

Label the probability axis as percentages (50, 60, ...) instead of probabilities (0.5, 0.6, ...)?

grid

Draw light gridlines?

mark

Optional numeric vector of differences in the predictor to highlight. For each one, a guide line runs up from the x-axis to the curve and the probability is labelled at the point. Negative values are only shown when full = TRUE.

mark_ci

Add the confidence interval to the labels of marked points?

title, subtitle

Plot title and subtitle. None by default; supply a string, or TRUE for an automatic one (the subtitle then describes the method, correlation, sample size and band).

...

Passed from plot() to autoplot().

Value

autoplot() returns a ggplot object; plot() returns it invisibly.

Examples

fit <- posc_uni(r = 0.5, n = 120, sd_x = 15,
                predictor_name = "IQ", outcome_name = "job performance")
plot(fit)

plot(fit, full = TRUE)

plot(fit, mark = c(10, 20, 30))

plot(fit, mark = 15, mark_ci = TRUE, percent = TRUE)

plot(fit, title = TRUE, subtitle = TRUE)