R: Response Curve Plots for IRT Models

curveplot {psychotools}

R Documentation

Response Curve Plots for IRT Models

Description

Base graphics plotting function for response curve plot visualization of IRT models.

Usage

  curveplot(object, ref = NULL, items = NULL, names = NULL,
    layout = NULL, xlim = NULL, ylim = c(0, 1), col = NULL,
    lty = NULL, main = NULL, xlab = "Latent trait",
    ylab = "Probability", add = FALSE, ...)

Arguments

`object`	a fitted model object of class `"raschmodel"`, `"rsmodel"` or `"pcmodel"`.
`ref`	argument passed over to internal calls of `predict`.
`items`	character or numeric, specifying the items for which response curves should be visualized.
`names`	character, specifying labels for the items.
`layout`	matrix, specifying how the response curve plots of different items should be arranged.
`xlim, ylim`	numeric, specifying the x and y axis limits.
`col`	character, specifying the colors of the response curve lines. The length of `col` should be the maximum number of available categories.
`lty`	numeric, specifying the line type of the response curve lines. The length of `lty` should either be one or the maximum number of available categories. In the first case, a single line type is used for all category response curves. In the latter case, separate line types for each category response curve are used.
`main`	character, specifying the overall title of the plot.
`xlab, ylab`	character, specifying the x and y axis labels.
`add`	logical. If `TRUE`, new response curves are added to an existing plot. Only possible when a single item is visualized.
`...`	further arguments passed to internal calls of `matplot`

Details

The response curve plot visualization illustrates the predicted probabilities as function of the ability parameter θ under a certain IRT model. This type of visualization is sometimes also called item/category operating curves or item/category characteristic curves.

Examples

## Load Verbal Aggression data
data("VerbalAggression", package = "psychotools")

## Fit Rasch, rating scale and partial credit
## model to VerbalAggression data
rmmod <- raschmodel(VerbalAggression$resp2)
rsmod <- rsmodel(VerbalAggression$resp)
pcmod <- pcmodel(VerbalAggression$resp)

## Curve plots of the dichotomous RM
plot(rmmod, type = "curves")

## Curve plots under the rating scale model
## for the first six items of the data set
plot(rsmod, type = "curves", items = 1:6)

## Curve plots under the partial credit model
## for the first six items of the data set
## with custom labels
plot(pcmod, type = "curves", items = 1:6,
  names = paste("Item", 1:6))

## Compare the predicted probabilities under the rating
## scale model and the partial credit model for a single item
plot(rsmod, type = "curves", item = 1)
plot(pcmod, type = "curves", item = 1, lty = 2, add = TRUE)
legend(x = "topleft", y = 1.0, legend = c("RSM", "PCM"), lty = 1:2, bty = "n")