SBIR Phase I: Authoring tools for rapid high quality complex assessments for Next Generation Science Standards

Siminsights Inc, Irvine CA

Investigators

Abstract

This SBIR Phase 1 project will research the feasibility of novel tools for authoring; distributing and scoring high quality game based complex performance assessments for Next Generation Science Standards, ultimately targeting the entire K-12 science segment of nearly 54 million students in the US. US science education is facing critical challenges in terms of global competitiveness. Designed to meet these challenges, Next Generation Science Standards call for major changes in assessments. Historically, the development life cycle of assessments, particularly for measuring complex learning, has required significant amount of time and resources. In addition, the design, development, and validation of complex measures of content and cognition can be expensive and take many months. Similarly, creation of simulations and games is also very expensive. Software systems that enable the authoring of large numbers of high quality game based assessments will accelerate assessment development and innovation, reduce costs, enhance teacher proficiency and buy-in, and also positively impact instruction and curriculum by providing valuable feedback on student learning progression. Highly engaging assessments will potentially help students realize where they need to focus their efforts without the pressure of typical tests, thereby increasing learning and self-efficacy outcomes and significantly increasing the number of students who decide to pursue science and engineering degrees and careers. This SBIR proposal seeks to simultaneously satisfy several important assessment requirements including quality, development cost, student engagement, easy authoring and teacher buy-in by creating a novel game based assessment authoring system. The proposed approach will investigate a middle path between two alternatives. The traditional approach involves assessment item development by professionals. While the resulting items are of high quality, they are also expensive and lack teacher buy-in and student engagement. In contrast, another approach calls for teachers to develop and test new items without much expert guidance. These items, while authentic and low-cost, can have quality problems. The proposed research will investigate authoring tools that offer teachers the freedom to exercise their creativity in authoring authentic assessments, while also providing expert guidance to significantly increase the probability that the assessments are of high quality and generate trustworthy formative feedback. The proposed system will also automatically measure and score students' performances using techniques from psychometrics and educational data mining to rapidly produce actionable insights to guide the next best instructional steps. The proposed authoring tools and resulting assessments will be validated with usability, classroom and crowdsourcing studies.