WrightMap Tutorial – Part 2

March 24, 2014
By

(This article was first published on R Snippets for IRT, and kindly contributed to R-bloggers)

Plotting Multidimensional & Polytomous Models


Remember: you can find the other parts of the tutorial here:


Multidimensional models

In Part 1, we reviewed how to install the package from GitHub and how to customize unidimensional and dichotomous models. Now in Part 2, we’ll look at graphing some more complicated models.

(See Part 3 for ConQuest integration and making thresholds out of deltas.)

Of course, you will need to load WrightMap to begin with as shown in Part1. We will also need to load again RColorBrewer for some of the examples.

library(RColorBrewer)

First, let’s generate some thresholds for a multidimensional model. This will be a matrix containing five columns of person estimates.

mdim.sim.thetas <- matrix(rnorm(5000), ncol = 5)

Since this will be a dichotomous model, we’ll generate a single column for thresholds.

mdim.sim.thresholds <- runif(10, -3, 3)

Okay, let’s see what the Wright Map looks like for this.

wrightMap(mdim.sim.thetas, mdim.sim.thresholds)

image

That doesn’t look right. Let’s adjust the proportion of the map’s parts.

wrightMap(mdim.sim.thetas, mdim.sim.thresholds, item.prop = 0.5)

image

Let’s change the dimensions names

wrightMap(mdim.sim.thetas, mdim.sim.thresholds, item.prop = 0.5, dim.names = c("Algebra", 
    "Calculus", "Trig", "Stats", "Arithmetic"))

image

And let’s give them some color

wrightMap(mdim.sim.thetas, mdim.sim.thresholds, item.prop = 0.5, dim.names = c("Algebra", 
    "Calculus", "Trig", "Stats", "Arithmetic"), dim.color = brewer.pal(5, "Set1"))

image

And let’s associate the items with each dimension

wrightMap(mdim.sim.thetas, mdim.sim.thresholds, item.prop = 0.5, dim.names = c("Algebra", 
    "Calculus", "Trig", "Stats", "Arithmetic"), dim.color = brewer.pal(5, "Set1"), 
    show.thr.lab = FALSE, thr.sym.col.fg = rep(brewer.pal(5, "Set1"), each = 2), 
    thr.sym.col.bg = rep(brewer.pal(5, "Set1"), each = 2), thr.sym.cex = 2, 
    use.hist = FALSE)

image

Polytomous models

All right, let’s look at a Rating Scale Model. First, let’s generate three dimensions of person estimates.

rsm.sim.thetas <- data.frame(d1 = rnorm(1000, mean = -0.5, sd = 1), d2 = rnorm(1000, 
    mean = 0, sd = 1), d3 = rnorm(1000, mean = +0.5, sd = 1))

Now let’s generate the thresholds for the polytomous items. We’ll make them a matrix where each row is an item and each column a level.

items.loc <- sort(rnorm(10))

rsm.sim.thresholds <- data.frame(l1 = items.loc - 1, l2 = items.loc - 0.5, l3 = items.loc + 
    0.5, l4 = items.loc + 1)

rsm.sim.thresholds
##          l1       l2      l3      l4
## 1  -3.19202 -2.69202 -1.6920 -1.1920
## 2  -2.55843 -2.05843 -1.0584 -0.5584
## 3  -1.37052 -0.87052  0.1295  0.6295
## 4  -1.25838 -0.75838  0.2416  0.7416
## 5  -1.10905 -0.60905  0.3909  0.8909
## 6  -1.09854 -0.59854  0.4015  0.9015
## 7  -0.97131 -0.47131  0.5287  1.0287
## 8  -0.94265 -0.44265  0.5573  1.0573
## 9  -0.46542  0.03458  1.0346  1.5346
## 10 -0.04582  0.45418  1.4542  1.9542

Let’s look at the Wright Map!

wrightMap(rsm.sim.thetas, rsm.sim.thresholds)

image

Let’s assign a color for each level

itemlevelcolors <- matrix(rep(brewer.pal(4, "Set1"), 10), byrow = TRUE, ncol = 4)

itemlevelcolors
##       [,1]      [,2]      [,3]      [,4]     
##  [1,] "#E41A1C" "#377EB8" "#4DAF4A" "#984EA3"
##  [2,] "#E41A1C" "#377EB8" "#4DAF4A" "#984EA3"
##  [3,] "#E41A1C" "#377EB8" "#4DAF4A" "#984EA3"
##  [4,] "#E41A1C" "#377EB8" "#4DAF4A" "#984EA3"
##  [5,] "#E41A1C" "#377EB8" "#4DAF4A" "#984EA3"
##  [6,] "#E41A1C" "#377EB8" "#4DAF4A" "#984EA3"
##  [7,] "#E41A1C" "#377EB8" "#4DAF4A" "#984EA3"
##  [8,] "#E41A1C" "#377EB8" "#4DAF4A" "#984EA3"
##  [9,] "#E41A1C" "#377EB8" "#4DAF4A" "#984EA3"
## [10,] "#E41A1C" "#377EB8" "#4DAF4A" "#984EA3"

And now make a Wright Map with them

wrightMap(rsm.sim.thetas, rsm.sim.thresholds, thr.sym.col.fg = itemlevelcolors, 
    thr.sym.col.bg = itemlevelcolors)

image

But we also want to indicate which dimension they belong… with symbols

itemdimsymbols <- matrix(c(rep(16, 12), rep(17, 12), rep(18, 16)), byrow = TRUE, 
    ncol = 4)

itemdimsymbols
##       [,1] [,2] [,3] [,4]
##  [1,]   16   16   16   16
##  [2,]   16   16   16   16
##  [3,]   16   16   16   16
##  [4,]   17   17   17   17
##  [5,]   17   17   17   17
##  [6,]   17   17   17   17
##  [7,]   18   18   18   18
##  [8,]   18   18   18   18
##  [9,]   18   18   18   18
## [10,]   18   18   18   18
wrightMap(rsm.sim.thetas, rsm.sim.thresholds, show.thr.lab = FALSE, thr.sym.col.fg = itemlevelcolors, 
    thr.sym.col.bg = itemlevelcolors, thr.sym.pch = itemdimsymbols, thr.sym.cex = 2)

image

Changing the item axis

What happens if you have too many items?

rasch2.sim.thresholds <- runif(50, -3, 3)

We use the defaults…

wrightMap(rnorm(1000), rasch2.sim.thresholds)

image

Some options…

wrightMap(rnorm(1000), rasch2.sim.thresholds, show.thr.lab = FALSE, label.items.srt = 45)

image

wrightMap(rnorm(1000), rasch2.sim.thresholds, show.thr.lab = FALSE, label.items.rows = 2)

image

wrightMap(rnorm(1000), rasch2.sim.thresholds, show.thr.lab = FALSE, label.items = c("", 
    "example", 1:48), label.items.rows = 3)

image

Or you can get rid of that axis completely

wrightMap(rnorm(1000), rasch2.sim.thresholds, show.thr.sym = FALSE, thr.lab.text = paste("I", 
    1:50, sep = ""), label.items = "", label.items.ticks = FALSE)

image

To leave a comment for the author, please follow the link and comment on his blog: R Snippets for IRT.

R-bloggers.com offers daily e-mail updates about R news and tutorials on topics such as: visualization (ggplot2, Boxplots, maps, animation), programming (RStudio, Sweave, LaTeX, SQL, Eclipse, git, hadoop, Web Scraping) statistics (regression, PCA, time series, trading) and more...



If you got this far, why not subscribe for updates from the site? Choose your flavor: e-mail, twitter, RSS, or facebook...

Comments are closed.