make.pseudo: Transform ratings from real raters into pseudo ratings

In sachseka/eatPrep: Prepare Data for IRT Analyses

Transform ratings from real raters into pseudo ratings

Description

Data from large-scale assessments often are rated by multiple raters. This function reduces the number of raters by the use of “pseudo raters”.

Usage

make.pseudo(datLong, idCol, varCol, codCol, alwaysPrefer = NULL,
            alwaysNeglect = NULL, valueCol, n.pseudo, item.groups = NULL,
            randomize.order = TRUE, verbose = FALSE)

Arguments

datLong	Data set in the longformat, i.e. one row per examinee-variable-rater combination.
idCol	Name or column number of the person identifier (ID) variable.
varCol	Name or column number of the variable identifier.
codCol	Name or column number of the rater identifier variable.
alwaysPrefer	Optional: Single name of the rater that should always be preferred if it is part of a double coding.
alwaysNeglect	Optional: Single name of the rater that should always be neglected if it is part of a double coding. This means that coding by this rater is only taken into account if no other rater has rated this response.
valueCol	Name or column number of the value variable.
n.pseudo	How many pseudo raters should be used? (value must be lower than the number of real raters)
item.groups	Optional: List of linked variables that could later be aggregated into items. If two raters make inconsistent judgments, it may be more favorable for aggregation if all variables belonging to an item are coded by the same rater per examinee. However, if a rater has not rated all the variables of the item, the ratings of other raters are used for the variables not rated by this rater.
randomize.order	Logical: if TRUE, the selection of raters to pseudo raters is random.
verbose	Logical: give information about number of persons, variables, raters?

Value

A data.frame in the long format.

Author(s)

Sebastian Weirich

Examples

data(rater)
oneRater <- make.pseudo (datLong=rater, idCol="id", varCol="variable", codCol="rater",
            valueCol="value", n.pseudo=1, verbose=TRUE)
twoRaters<- make.pseudo (datLong=rater, idCol="id", varCol="variable", codCol="rater",
            valueCol="value", n.pseudo=2)
# with item groups, allowing that all variables belonging to an item are coded by the
# same rater (per examinee)
itemGroup<- list(first = c("V01", "V03"), second = c("V05", "V06", "V07"))
oneRater2<- make.pseudo (datLong=rater, idCol="id", varCol="variable", codCol="rater",
            item.groups = itemGroup, valueCol="value", verbose=TRUE, n.pseudo=1)
attr(oneRater2, "info")
# prefer John
oneRater <- make.pseudo (datLong=rater, idCol="id", varCol="variable", codCol="rater",
            alwaysPrefer = "John", valueCol="value", n.pseudo=1, verbose=TRUE)
# neglect Edward
oneRater <- make.pseudo (datLong=rater, idCol="id", varCol="variable", codCol="rater",
            alwaysNeglect = "Edward", valueCol="value", n.pseudo=1, verbose=TRUE)
# prefer Dolores and neglect Edward
oneRater <- make.pseudo (datLong=rater, idCol="id", varCol="variable", codCol="rater",
            alwaysPrefer = "Dolores", alwaysNeglect = "Edward", valueCol="value",
            n.pseudo=1, verbose=TRUE)

sachseka/eatPrep documentation built on June 9, 2025, 9:36 a.m.