Thesis advisor: Michael Russell / Large-scale assessments are often used for statewide accountability and for instructional and institutional planning. It is essential that the instruments used are valid and reliable for all test takers included in the testing population. However, these tests have often fallen short in the area of accessibility, which can impact validity for students with special needs. This dissertation examines two strategies to addressing accessibility: the use of technology to implement principles of universal design to assessment and the provision of accommodations. This study analyzed test data for students attending high schools in New Hampshire, Vermont and Rhode Island who participated in the 2009 11th grade New England Common Assessment Program (NECAP) science assessment. Three test conditions were of interest: (1) no accommodations with a paper-based form (2) accommodated test administration with a paper-based form and (3) accommodated test administration using a universally designed computer-based test delivery system with embedded accommodations and accessibility features. Results from two analyses are presented: differential item functioning (DIF) and confirmatory factor analysis (CFA). DIF was used to explore item functioning, comparing item difficulty and discrimination under accommodated and non-accommodated conditions. Similarly, CFA was used to examine the consistency of underlying factor structure as evidence that constructs measured were stable across test conditions. Results from this study offered evidence that overall item functioning and underlying factor structure was consistent across accommodated and unaccommodated conditions, regardless of whether accommodations were provided with a paper form or a universally designed computer-based test delivery system. These results support the viability of using technology-based assessments as a valid means of assessing students and offering embedded, standardized supports to address access needs.

La enseñanza y evaluación automática a través de un sistema Computer Based Assessment (CBA) requiere de software especializado que se adapte a la tipología de actividades a tratar y evaluar. En esta tesis se ha desarrollado un entorno CBA que facilita el aprendizaje y evaluación de los principales temas de una asignatura de bases de datos. Para ello se han analizado las herramientas existentes en cada uno de estos temas (Diagramas Entidad/Relación, diagramas de clases, esquemas de bases de datos relacionales, normalización, consultas en álgebra relacional y lenguaje SQL) y para cada uno de ellos se ha analizado, diseñado e implementado un módulo de corrección y evaluación automática que aporta mejoras respecto a los existentes. Estos módulos se han integrado en un mismo entorno al que hemos llamado ACME-DB. / Teaching and automatic evaluation through Computer Based Assessment (CBA) requires of specialized software tools capable to correct different type of activities. In this work, we have analyzed the main topics of a database course (entity/ relationship diagrams, class diagrams, relational database schemas, normalization, relational algebra and SQL) and for each one of them we have reported the state of art on CBA tools. Afterwards, we have proposed a new CBA environment for teaching and learning database. This environment, denoted ACME-DB, is composed of different modules capable to automatically correct and evaluate exercises of main topics of a database course providing improvements over existing tools.