drm-methods: Dichotomous Response Model Probabilities

Description

This function computes the probability of correct responses (and optionally, incorrect responses) for one or more items for a given set of theta values using the 1PL, 2PL, 3PL, M1PL, M2PL or M3PL model, depending on the included item parameters and the specified number of dimensions.

Usage

drm(x, theta, dimensions = 1, D = 1, incorrect = FALSE, 
  print.mod = FALSE, items, information = FALSE, angle, ...)
## S4 method for signature 'numeric'
drm(x, theta, dimensions, D, incorrect, print.mod, items, information, angle, ...)
## S4 method for signature 'matrix'
drm(x, theta, dimensions, D, incorrect, print.mod, items, information, angle, ...)
## S4 method for signature 'data.frame'
drm(x, theta, dimensions, D, incorrect, print.mod, items, information, angle, ...)
## S4 method for signature 'list'
drm(x, theta, dimensions, D, incorrect, print.mod, items, information, angle, ...)
## S4 method for signature 'irt.pars'
drm(x, theta, dimensions, D, incorrect, print.mod, items, information, angle, ...)
## S4 method for signature 'sep.pars'
drm(x, theta, dimensions, D, incorrect, print.mod, items, information, angle, ...)

Arguments

an R object containing item parameters

theta

vector, matrix, or list of theta values for which probabilities will be computed. If theta is not specified, an equal interval range of values from -4 to 4 is used with an increment of 0.5. See details below for more information.

dimensions

number of modeled dimensions

scaling constant. The default value assumes that the parameters are already in the desired metric. If the parameters are in the logistic metric, they can be transformed to a normal metric by setting D = 1.7

incorrect

if TRUE, compute probabilities for the incorrect response category)

print.mod

if TRUE, print the model that was used (i.e., 1PL, 2PL, 3PL, M1PL, M2PL, or M3PL)

items

numeric vector identifying the items for which probabilities should be computed

information

logical value. If TRUE compute item information. In the multidimensional case, information will be computed in the directions specified by angle or default angles of 0 - 90 in increments of 10 degrees.

angle

vector or matrix of angles between the dimension 1 axis and the corresponding axes for each of the other dimensions for which information will be computed. When there are more than two dimensions and angle is a vector, the same set of angles will be used relative to each of the corresponding axes.

...

further arguments passed to or from other methods

Value

Returns an object of class

Methods

x = "numeric"

This method should only be used for the Rasch model where x is a vector of difficulty parameters (or parameters related to item difficulty in the multidimensional case). Under this method the slopes and lower asymptote values for all items will default to one and zero respectively. This is true for both the unidimensional and multidimensional case.

x = "matrix"

This method allows one to specify an n x k matrix for n items, for m dimensions, and k equal to m, m+1, or m+2. The first m columns generally correspond to the discrimination/slope parameters, and the last two columns correspond to the difficulty, and lower asymptote parameters, although this may not be the case depending on the supplied parameters.

1PL/M1PL:: For the 1PL and M1PL model with discriminations equal to 1 (Rasch Model), an n x 1 matrix of item difficulties can be supplied. An n x (m+1) matrix can also be used with all the values in the first m columns equal to 1 and difficulty parameters in the last column. For discrimination values other than 1, x should include at least m+1 columns where the first m columns contain the item discriminations (identical for all items) and the last column is for the item difficulties. The lower asymptote defaults to zero for all items; however, m+2 columns can be specified where the values in the last column all equal zero.
2PL/M2PL:: x should include at least m+1 columns where the first m columns contain the item discriminations and the last column is for the item difficulties. The lower asymptote defaults to zero for all items; however, m+2 columns can be specified where the values in the last column all equal zero.
3PL/M3PL:: x should include m+2 columns where the first m columns contain the item discriminations, the m+1 column is for item difficulties, and the last column is for lower asymptote values.

x = "data.frame"

See the method for x = "matrix"

x = "list"

This method can include a list with 1, 2, or 3 elements. In general, these elements correspond to discrimination, difficulty, and lower asymptote parameters, although this may not be the case depending on the supplied parameters. For the unidimensional case, each list element can be a vector of length n or an n x 1 matrix for n items. For the multidimensional case, the list element for the discrimination parameters should be an n x m matrix for m dimensions. The other list elements can be vectors of length n or an n x 1 matrix for n items.

1PL/M1PL:: For the 1PL and M1PL model with discriminations equal to 1 (Rasch Model), one element with item difficulties can be supplied. Alternatively, two elements can be used with the first list element containing a matrix/vector of ones and difficulty parameters in the second list element. For discrimination values other than 1, x should contain at least two list elements where the first contains the item discriminations (identical for all items) and the second is for item difficulties. The lower asymptote defaults to zero for all items; however, a third element with a vector/matrix of zeros can be included.
2PL/M2PL:: x should contain at least two list elements where the first element contains the item discriminations and the second element includes the item difficulties. The lower asymptote defaults to zero for all items; however, a third element with a vector/matrix of zeros can be included.
3PL/M3PL:: x should include three list elements where the first element contains the item discriminations, the second element includes the item difficulties, and the third element contains the lower asymptote values.

x = "irt.pars"

This method can be used to compute probabilities for the dichotomous items in an object of class "irt.pars". If x contains polytomous items, a warning will be displayed stating that probabilities will be computed for the dichotomous items only. If x contains parameters for multiple groups, a list of "irt.prob" objects will be returned. The argument dimensions does not need to be included for this method.

x = "sep.pars"

This method can be used to compute probabilities for the dichotomous items in an object of class sep.pars. If x contains polytomous items, a warning will be displayed stating that probabilities will be computed for the dichotomous items only. The argument dimensions does not need to be included for this method.

Details

theta can be specified as a vector, matrix, or list. For the unidimensional case, theta should be a vector. If a matrix or list of values is supplied, they will be converted to a single vector of theta values. For the multidimensional case, if a vector of values is supplied it will be assumed that this same set of values should be used for each dimension. Probabilities will be computed for each combination of theta values. Similarly, if a list is supplied, probabilities will be computed for each combination of theta values. In instances where probabilities are desired for specific combinations of theta values, a j x m matrix should be specified for j ability points and m dimensions where the columns are ordered from dimension 1 to m.

References

Adams, R. J., Wilson, M., & Wang, W. (1997). The multidimensional random coefficients multinomial logit model. Applied Psychological Measurement, 21(1), 1-23. Birnbaum, A. (1968). Some latent trait models and their use in inferring an examinee's ability In F. M. Lord & M. R. Novick (Eds.) Statistical Theories of Mental Test Scores. Reading, MA: Addison-Wesley. Rasch, G. (1960). Probabilistic Models for Some Intelligence and Attainment Tests Copenhagen, Denmark: Danish Institute for Educational Research. Reckase, M. D. (1985). The difficulty of items that measure more than one ability. Applied Psychological Measurement, 9(4), 401-412. Reckase, M. D. (1997). A linear logistic multidimensional model for dichotomous item response data. In W. J. van der Linden & R. K. Hambleton (Eds.) Handbook of Modern Item Response Theory (pp. 271-286). New York: Springer-Verlag. Weeks, J. P. (2010) plink: An R package for linking mixed-format tests using IRT-based methods. Journal of Statistical Software, 35(12), 1--33. URL http://www.jstatsoft.org/v35/i12/

Examples

Run this code

###### Unidimensional Examples ######
## 1PL
# A single item with a difficulty at 1 logit
x <- drm(1) 
plot(x)

# Three items with a difficulties at -2.2, -1.7, and 0.54 logits 
# respectively. Assume a constant discrimination of 1 for each item
b <- c(-2.2,-.7,1.54)
x <- drm(b) 
# Plot the expected probabilites
plot(x)

# Five items with a constant discrimination equal to 1.2
# Use the scaling constant for the normal metric
a <- rep(1.2,5)
b <- c(-2,-1,0,1,2)
pars <- list(a,b)
x <- drm(pars,D=1.7)
# Extract the expected probabilities
get.prob(x)
# Plot the expected probabilities
plot(x,combine=5,item.names="Items 1-5")


## 2PL
# Compute probabilities for five items at a theta value of 1.5 logits
# Print the model
a <- c(1,.8,.6,1.2,.9)
b <- c(-1.7,2.2,.4,-1,-.4)
pars <- cbind(a,b)
drm(pars,theta=1.5,print.mod=TRUE)


# Include a vector of zeros for the lower asymptote
# Compute probabilities for incorrect responses
a <- c(.63,1.14,.89,1.01,1.51,.58)
b <- c(-1.9,.08,1.6,-1.4,.5,-2.3)
c <- rep(0,6)
pars <- cbind(a,b,c)
x <- drm(pars,incorrect=TRUE)
plot(x)


## 3PL
a <- c(1,.8,.4,1.5,.9)
b <- c(-2,-1,0,1,2)
c <- c(.2,.25,.18,.2,.22)
pars <- list(a,b,c)
x <- drm(pars)
plot(x)

# Use theta values from -3 to 3 with an increment of 0.2
a <- c(.71,.96,.36,1.05,1.76,.64)
b <- c(-.16,1.18,2.03,1.09,.82,-1.56)
c <- c(.22,.17,.24,.08,.20,.13)
theta <- seq(-3,3,.2)
pars <- cbind(a,b,c)
x <- drm(pars,theta)
plot(x,combine=6,item.names="Items 1-6",auto.key=list(space="right"))

###### Multidimensional Examples ######
## M1PL
# A single item with a parameter related to item difficulty at 1 logit
x <- drm(1, dimensions=2) 
plot(x)

# Three items with a difficulties at -2.2, -1.7, and 0.54 logits 
# respectively. Assume a constant discrimination of 1 for each item
d <- c(-2.2,-.7,1.54)
x <- drm(d, dimensions=2) 
plot(x, drape=TRUE)


## M2PL
# Items 27-30 from Reckase (1985)
a <- matrix(c(1.66,1.72,.69,.19,.88,1.12,.68,1.21),4,2,byrow=TRUE)
d <- c(-.38,-.68,-.91,-1.08)
pars <- list(a,d)
x <- drm(pars,dimensions=2)
plot(x, drape=TRUE, item.names=paste("Item",27:30))

# Create contourplots for these items
plot(x,type="contourplot",cuts=10)


## M3PL
# Single item from Reckase (1997, p. 274)
pars <- t(c(.8,1.4,-2,.2))
x <- drm(pars, dimensions=2)
plot(x, default.scales=list(arrows=FALSE),drape=TRUE)

# Compute the probabilities of an incorrect response
# for the Reckase (1997) item
x <- drm(pars, dimensions=2, incorrect=TRUE)
plot(x, screen=list(z=-40,x=-60), auto.key=list(space="right"))

# Four items from the included example for BMIRT (Yao, 2003)
# modeled using four dimensions
pars <- matrix(c(0.5038,2.1910,1.1317,0.2493,0.5240,0.1763, 
0.2252,1.1999,0.5997,0.2087, 2.1730,0.4576,
0.2167,0.2487,1.4009,0.3865,-1.5270,0.3507, 
2.3428,1.1530,0.3577,0.4240,-1.4971,0.3641),4,6,byrow=TRUE)
colnames(pars) <- c("a1","a2","a3","a4","d","c")
x <- drm(pars, dimensions=4, print.mod=TRUE)


## Not run: ------------------------------------
# # Plot the item response surfaces for item 4
# plot(x,items=4,item.names="Item 4",drape=TRUE)
## ---------------------------------------------

Run the code above in your browser using DataLab