Validity and reliability of an unbounded item bank for placement and diagnostics in engineering.

Abstract

In the admission process for engineering faculties, the predominant selection method is evaluation through multiple-choice items. Under the premise that high scores positively correlate with academic success, it is imperative to have robust and reliable instruments. However, given the evolution towards Society 4.0 and 5.0, these processes must integrate advanced technologies and inclusive approaches that mitigate discrimination. This paper presents the initial phase of the MicroCell project, aimed at defining preparatory strategies for Education 2030, focused on reducing academic deficiencies detected through the selection process. The study centers on the design and validation of an assessment model based on an unbounded item bank, analyzing its quality and robustness as a placement tool. The results demonstrate that this format not only maintains the technical integrity of the exam but also effectively neutralizes random response bias. Thus, the proposal serves not only for selection purposes but also as a diagnostic tool to implement personalized academic actions that enhance the future engineer’s performance.