Title: eatATA: Automated Test Assembly for R-Users

Authors: Dries Debeer, Benjamin Becker

Abstract:
In psychological and educational assessment, combining items from an existing
item pool into new test forms is a well-known problem, generally referred to
as Test Assembly. Especially when multiple test forms need to be assembled,
while also maintaining psychometric and item-content related requirements in
each form, these problems can quickly become challenging. Solving such a
complex combinatorial optimization problem by hand is inefficient, error
prone, and likely to result in sub-optimal solutions. Yet, because these
assembly problems can be viewed as (and translating into) mixed integer linear
programming problems, they can be efficiently addressed through software tools
with high-end mathematical programming algorithms. However, although these
software tools can be used to perform assembly tasks efficiently and
automatically, they were generally not designed with automated test assembly
in mind. Consequently, these tools are often unknown or unavailable to the
researchers that are confronted with test assembly problems. The R-package
eatATA aims at making automated test assembly for R users as easy as fitting a
linear regression model. The package was designed specifically from the
perspective of Automated Test Assembly (ATA), and makes several software tools
for mixed-integer linear programming available in a straightforward and
convenient manner. In this presentation, we discuss the general functionality
of the eatATA package and, using practical use cases, we demonstrate how its
workflow is intuitive for R-users.