item.correlations: Item correlations

In jeppebundsgaard/RDigram: Analysis of Items from Tests and Surveys

Item correlations

Description

Calculate item correlations, item-rest correlations and correlations between items and exogenous variables

Usage

item.correlations(do=NULL,resp=NULL,items=1:do$recursive.structure[1],exo=(do$recursive.structure[1]+1):do$recursive.structure[2])

Arguments

do	an object of class digram.object
resp	A data.frame or matrix of recoded data (only used if do is NULL)
items	A vector of columns from the recoded data to include as items in the analysis or a character vector of variable labels
do.testlets	Bolean. If TRUE, testlets are combined to superitems.
do.split	Bolean. If TRUE, items coded as split are split.
exo	A vector of columns from the recoded data to include as exogenous variables in the analysis or a character vector of variable labels
max.name.length	Maximum length of item names (to be printed in tables)
accept.na	A boolean. Include cases with missing values in responses
verbose	Print results
extra.verbose	Print warnings in PDF and HTML-output
caption.items	Caption of items table
caption.rest	Caption of item-rest table
caption.exo	Caption of exogenous variables table

Details

First step in item screening: Analysis of consistency (Positive correlations)

M1

Y_i and Y_j are positively correlated for all pairs of items

M2

Y_a is positively monotonically related to the rest-score R_a and all subscores S_B where Y_a \not\in B

M3

If X is positively related to \theta, then X will also be positively related to S, to all subscores, S_A, and all item responses Y_i

Value

Returns a list of correlations

Author(s)

Jeppe Bundsgaard [e-mail]

References

Kreiner, S. & Christensen, K.B. (2011). Item Screening in Graphical Loglinear Rasch Models. Psychometrika, vol. 76, no. 2, pp. 228-256. DOI: 10.1007/s11336-9203-Y

Examples

library(iarm)
do<-digram.object(project = "desc2",data = desc2,variables = c(5:14,2:4,1),recursive.structure = c(10,13))
item.correlations(do)

jeppebundsgaard/RDigram documentation built on Oct. 29, 2023, 7:15 p.m.