Beyond mere novelty: Categorization drives differentiation.

Competence-based knowledge structure theory (CbKST; Doignon, ; Düntsch & Gediga, ; Korossy, ) allows the identification of the specific attributes (skills, symptoms) possessed by an individual based on their responses to test items. Assessments conducted within this framework span a variety of domains in which items can be meaningfully linked to the attributes required to solve or endorse them. Examples include hard skills (e.g., arithmetic, Anselmi, Stefanutti, de Chiusole, & Robusto, ; basic statistics, de Chiusole, Stefanutti, Anselmi, & Robusto, , ), soft skills (e.g., collaborative and teamwork skills, Morleo, Anselmi, & Vitanza, ; social and emotional skills, Morleo, Vitanza, & Anselmi, ), psychological functioning (e.g., depression, Serra, Spoto, Ghisi, & Vidotto, ), and neuropsychological functioning (e.g., fluid intelligence, de Chiusole et al., ). These assessments may also differ substantially in the number of attributes and items involved. For instance, CoTeSt (Collaborative and Teamwork Skill test) is a 20-item test aimed at assessing which of six collaborative and teamwork skills teachers possess (Morleo et al., ), whereas MatriKS 4–11 is an adaptive test consisting of 38 Raven-like matrices and aimed at assessing which of the nine skills required to solve them children possess (de Chiusole et al., ). The present work focuses on the educational assessment context. Accordingly, the attributes under consideration are referred to as “skills”, and the items are described as solved or not because responses are typically interpreted as correct or incorrect. In the clinical assessment context, the attributes may instead be referred to as “symptoms” or “diagnostic criteria”, and the items may be described as endorsed or not because responses are typically not interpreted in terms of correctness. The present work further considers conjunctive tests, in which an item is expected to be solved or endorsed if all the attributes associated with it are possessed. This type of test is especially appropriate in educational settings, where solving an item usually requires a series of steps, all of which must be carried out correctly. However, the proposed methodology also applies to the development of disjunctive tests, in which an item is expected to be solved or endorsed if at least one of the attributes associated with it is possessed. Disjunctive tests may be more suitable in some clinical contexts, where a disorder may be diagnosed based on the presence of only one of its symptoms (Anselmi, Heller, Stefanutti, & Robusto, ; Roussos, Templin, & Henson, ; Templin & Henson, ). Competence-based test development (CbTD; Anselmi, Heller, Stefanutti, & Robusto, , ) is a recent and innovative method for developing tests that are as informative as possible about the specific skills that an individual possesses. If desired, the tests can also be minimal, meaning that the exclusion of any item would reduce their informativeness. CbTD can be used in various test development contexts, including constructing tests from scratch, improving existing tests, and shortening them. For example, Anselmi, Heller, Stefanutti, Robusto, and Barillari () constructed a test for the assessment of students’ mathematical formulation, improved a booklet from the TIMSS (Trends in International Mathematics and Science Study; ) 2003 eighth-grade mathematics test, and shortened Tatsuoka ’s () fraction subtraction test. Dörrbecker and Heller () shortened the subtraction subtest of the Heidelberger Rechentest 1–4 (Haffner, Baro, Parzer, & Resch, ), which is one of the most commonly used German elementary school speed numeracy tests.