checkLinking: check for linked design

In weirichs/eatModel: Educational Assessment Tools for IRT Modeling using Conquest or TAM

check for linked design

Description

Function checks whether all blocks in a complete or incomplete block design are linked to each other.

Usage

checkLinking(design, blocks=NULL, bookletColumn=NULL, verbose=FALSE) 

Arguments

design	A data frame with the test design. All columns (except for the booklet identifier column) are expected to contain blocks.
blocks	Optional. To check whether a subdomain is completely linked add here all the blocks that belong to this subdomain in a character vector.
bookletColumn	Optional. Number or name of the booklet identifier column in the design data frame.
verbose	Optional. If TRUE the function gives more messages.

Value

A list. Containing information about linking diagnostics:

completelyLinked	A logical value (TRUE/FALSE). Whether design is completely linked.
occuringBlockCombinations	A data.frame containing the occureng frequencies of block pairs.
blockPositions	A data.frame containing information about how often each block occurs at each position.

Author(s)

Sebastian Weirich and Karoline Sachse

Examples

# 1. first examples
# a) design linked
des1a <- data.frame(booklet = paste0("B", 1:4),
       Pos1 = c("blockA", "blockB", "blockC", "blockD"),
       Pos2 = c("blockB", "blockC", "blockD", "blockE"),
       Pos3 = c("blockC", "blockD", "blockE", "blockF"))
test1 <- checkLinking(design = des1a, bookletColumn = "booklet", verbose=TRUE)

# b) design not linked:
des1b <- data.frame(Pos1 = c("blockA", "blockH", "blockC", "blockF"),
       Pos2 = c("blockB", "blockC", "blockD", "blockA"),
       Pos3 = c("blockF", "blockG", "blockH", "blockB"))
test2 <- checkLinking(design = des1b, verbose=TRUE)

# 2. second examples: use a 'IQB Bildungstrend 2016'-like design
data(des2)

# design contains three dimensions -- reading, listening and orthography
# linking check must be conducted for each dimension separately.

# a) domain reading (blocks contain "-L")
readblocks <- grep("-L", unique(unlist(des2[,-1])), value=TRUE)
test3 <- checkLinking(design = des2, blocks =readblocks, bookletColumn = "TH", verbose=TRUE)

# b) domain listening (blocks contain "-H")
listenblocks <- grep("-H", unique(unlist(des2[,-1])), value=TRUE)
test4 <- checkLinking(design = des2[,-1], blocks =listenblocks, verbose=TRUE)

# c) domain orthography (blocks contain "-R")
orthoblocks <- grep("-R", unique(unlist(des2[,-1])), value=TRUE)
test5 <- checkLinking(design = des2[,-1], blocks =orthoblocks, verbose=TRUE)

# reconstruct test design from exemplary data, separately for each year and each domain
data(trends)
design<- by(data = trends, INDICES = trends[,c("year", "domain")], FUN = function (d) {
         rownames(d) <- NULL
         dw <- reshape2::dcast(unique(d[,c("booklet", "block", "pos")]), booklet~pos, value.var="block")
         message("\nCondition: \n", eatTools::print_and_capture(d[1,c("year", "domain")]))
         cl <- checkLinking(design=dw, bookletColumn ="booklet") } )

weirichs/eatModel documentation built on Jan. 26, 2025, 4:01 p.m.