Math Test-Taking Behavior of Students With Learning Disabilities: A Natural Language Processing Approach

• In Event: Sheraton Roundtable Session Five
  In Roundtable Session: Fostering Mathematical Teaching and Learning for Students With Disabilities (Table 11)

Abstract

This study used n-gram natural language modeling to identify math problem-solving be-haviors of students with learning disabilities (LD) and their typically developing (TD) peers by analyzing 2017 NAEP eighth-grade process data collected on a math fraction item. As compared with TD students, students with LD were less likely to have a short and correct action sequence; instead, they exhibited more hesitant and repetitive behavior. Students with LD were also less likely to review and check their answer after submitting the correct answer than their TD peers. Furthermore, students with LD spent more time than TD students to decide whether to clear an answer, to eliminate a wrong choice, and to choose the correct answer.

Authors

• Susu Zhang, University of Illinois at Urbana-Champaign
  Presenting Author

• Xin Wei, Digital Promise
  Non-Presenting Author