plot_or_heatmap: Odds ratio heatmap

In Famondir/birtms: Creates and analyses Bayesian IRT models

Description

Odds ratio values should be greater 1 for items that are conditionally independent in a one dimensional IRT model (Holland, Rosenbaum, 1986). Haberman (2007) shows that the lower bond can be estimated more precisely (and will be even higher than 1) and also a upper bond can be calculated and gives formulae for 1PL and 2PL models.

Usage

plot_or_heatmap(
  or_data,
  model = NULL,
  itemrange = NULL,
  a = 1,
  sigma = 1,
  median_centered_colorscale = FALSE,
  bayesian = FALSE,
  nsim_ci = 1e+06,
  ci_width = 0.89
)

Arguments

or_data	dataframe generated by birtms::get_or()
model	birtmsfit object
itemrange	integer vector of length 2; first and last item to incorporate in the heatmap
a	double; mean slope parameter (if not passed model)
sigma	double; standard deviation of person dimesnion (theta)
median_centered_colorscale	boolean
bayesian	boolean; should the bayesian distribution for actual odds ratio should be used as reference instead of point estimate
nsim_ci	integer; number of draws used to get a distribution of the odds ratio value to calculate CIs
ci_width	double

Value

ggplot2 object

Examples

## Not run: 
or_data <- get_or(fit, n_samples = 500)
plot_or_heatmap(or_data)

## End(Not run)

Famondir/birtms documentation built on Feb. 18, 2022, 2:51 a.m.