A recent AACR project focused on formative assessment in high school physical science was highlighted by the Community for Advancing Discovery Research in Education.  Check it out here:  https://cadrek12.org/spotlight/learning-progressions-trajectories#haudek

Our research work will be featured as part of a related paper set at the NARST 2024 Annual International Conference. The session is titled: Challenges of using AI for evaluation of knowledge-in-use assessments. The session focuses on some of the challenges and benefits of using AI as part of automated evaluation of complex, three-dimensional assessment tasks.  Come see what we've been working on!  

Dr. Leonora Kaldaras will present Rubric Development for AI Scoring of NGSS Learning Progression-Based Scientific Models To Support Individual Opportunity To Learn

Tingting Li will present Utilizing Deep Learning AI to Evaluate Scientific Models: Overcoming the Challenges

Conference information is here:https://narst.org/conferences/2024-annual-conference 

Dr. Kevin Haudek and Dr. Leonora Kaldaras presented research work as part of the 2023 DRK-12 PI Meeting hosted by CADRE (https://cadrek12.org).  As part of the NSF funded project, the team is developing an assessment system that can capture student text explanations and electronically drawn models. These student responses will be automatically scored using the Constructed Response Classifier developed by the AACR research group.  

As part of the session Using AI to Promote Equitable Science Teaching and Learning, Dr. Kaldaras presented on providing opportunity for three dimensional science learning for all students and how the project addresses equity in assessment items, coding rubrics and text scoring models. Later in the meeting, Dr. Haudek presented a poster describing the project progress to date, including the rubric development process.   

The 3DLP Auto Feedback project held a kickoff meeting on September 9, 2022 at the CREATE for STEM Institute at Michigan State to plan project work and discuss how to meet the project’s research goals. The meeting involved 9 participants from across colleges at MSU and University of Colorado.  Participants had research backgrounds in education, science, assessment and artificial intelligence.  During the meeting, participants reviewed project aims, identified barriers to year one progress and discussed connections between the proposed project work and ongoing investigations in similar fields.

The project is led by Drs. Kevin Haudek and Joe Krajcik at Michigan State and Dr. Leonora Kaldaras at University of Colorado-Boulder.

Automatic Feedback of Constructed Response Assessments is a four-year project funded by the National Science Foundation (Grant No. 2200757) that will study the effects of automated feedback provided to students on the Learning Progressions (LP)-aligned formative assessments along validated NGSS-aligned LP's. The project builds from previous successful research in automatic text scoring and 3D assessment tasks to build an assessment system that provides feedback to students. The automatic feedback tool paired with the NGSS-aligned LP will be immediately useful for many Interactions teachers and their students across the United States. The project is led by Drs. Kevin Haudek and Joe Krajcik at Michigan State and Dr. Leonora Kaldaras at University of Colorado-Boulder.