equi: Equipercentile equating

In twolodzko/equi: equi: Equipercentile Equating

Description

Functions for equipercentile equating in Equivalent Groups (EG), Single Group (SG) and Nonequivalent groups with Anchor Test using Chained Equating (NEAT-CE) designs.

Usage

equi(x, y, range, truncate = TRUE, df = "BIC", margin = 0.5)

## S3 method for class 'equi'
plot(x, diff = FALSE, ref = TRUE, type = "l", ...)

Arguments

x	smoothtab object or numeric vector. For NEAT first column of smoothtab data frame is source test and second is anchor test (see Details).
y	smoothtab or equi object. For NEAT first column of smoothtab data frame is anchor test and second is target test (see Details).
range	two-item vector of minimum and maximum score points of y.
truncate	if TRUE truncates the values using trun.
df	degrees of freedom for ns splines. If df is set to "BIC", "AIC" or "LSE" it uses BIC, AIC or sum of squared errors for choosing the best df value.
margin	see: trun.
diff	plot a difference from identity function.
ref	plot a reference line, i.e. an identity function.
type	type of the plot.
...	potentially further arguments passed from other methods.

Details

If x is smoothtab object and y is not provided, single group equating is applied. If x and y are smoothtab objects, equivalent groups or NEAT-CE equating is applied, depending on the kind of data provided. If x is numeric vector and y is equi object, x is transformed using equating function provided in y.

Value

Returns object of class equi, that can be used for transforming test scores using equi function (see examples).

References

Kolen, M.J. & Brennan, R.J. (2004). Test Equating, Scaling, and Linking: Methods and Practices. New York: Springer-Verlag.

von Davier, A.A., Holland, P.W. & Thayer, D.T. (2004). The Kernel Method of Test Equating. New York: Springer-Verlag.

Green, P.J. & Silverman, B.W. (1993). Nonparametric Regression and Generalized Linear Models: A roughness penalty approach. London: Chapman & Hall/CRC.

See Also

ns, bs, poly, approx

Examples

# Single Group design with presmoothing

data(Tests)
x <- Tests[Tests$Sample == "P", "x"]
y <- Tests[Tests$Sample == "P", "y"]

(eq <- equi(smoothtab(x, y, presmoothing=TRUE)))
yx <- equi(x, eq)

moments(list(x, y, yx))

# Equivalent Groups design

data(Tests)
x <- Tests[Tests$Sample == "P", "x"]
y <- Tests[Tests$Sample == "P", "y"]

(eq <- equi(smoothtab(x), smoothtab(y)))
yx <- equi(x, eq)

# NEAT-CE design, 'x' is equated to 'y' via anchor test 'a' (xa, ya).
# in 'p' first column is source test, second is anchor test
# in 'q' first column is anchor test, second is target test

data(Tests)
p <- Tests[Tests$Sample == "P", 1:2]
q <- Tests[Tests$Sample == "Q", 2:3]

(eq <- equi(smoothtab(p), smoothtab(q)))
yx <- equi(p[, 1], eq)

twolodzko/equi documentation built on Nov. 11, 2020, 4:04 p.m.