Simulated right/wrong responses to two forms of a fictional 40-item
licensure exam in a common-item nonequivalent groups (NEAT) design. Form A
was taken by 1,000 candidates in 2025 and Form B by 1,000 candidates in
2026. Each form has 28 unique items and 12 common anchor items. Problems
were planted on purpose for the linking lab;
lab_truth("linking") lists them.
A data frame with 2,000 rows and 70 columns: id (candidate ID),
form (factor, A or B), the Form A items A01 to A28, the Form B
items B01 to B28 and the anchors C01 to C12 (1 = right,
0 = wrong, NA = not on the candidate's form).
Source
Simulated; the script is data-raw/linking.R in the package's
GitHub repository.
Examples
table(linking$form)
#>
#> A B
#> 1000 1000
form_a <- linking[linking$form == "A", c(sprintf("A%02d", 1:28), sprintf("C%02d", 1:12))]
summary(rowSums(form_a))
#> Min. 1st Qu. Median Mean 3rd Qu. Max.
#> 0.00 14.00 22.00 20.86 27.00 40.00
