A method for generating nonverbal reasoning items using n-layer modeling

Abstract

Automatic item generation is the process of using item models to produce assessment tasks using computer technology. An item model is comparable to a template that highlights the variables or elements in the task that must be manipulated to produce new items. When a small number of elements is manipulated in the item model, the generated items look similar to one another and are often referred to as clones. The purpose of our study is to describe a method for generating large numbers of diverse and heterogeneous items using a generalized approach called n-layer item modeling. When a large numbers of elements is manipulated in the n-layer item model, diverse items are generated. We demonstrate the method by generating 1,340 nonverbal reasoning items that would be appropriate for a high-stakes medical admission test.