Analysis of online one-on-one math tutoring dialogues reveals distinct interaction patterns across grade levels, with high school students showing higher engagement and cognitive abilities while primary students benefit most from active tutor support.
Objective: To analyze dialogic interactions in online one-on-one mathematics tutoring across different educational levels using AI techniques and computational analysis.
Methods:
- Developed coding scheme for analyzing online tutoring dialogues
- Used AI techniques to build automated dialog annotation model
- Analyzed dataset from Singapore educational system covering grades 1-12
- Applied statistical methods and lag sequential analysis
Key Findings:
- High school students showed more active participation and superior reasoning skills
- Primary students were less active but responded positively when engaged by tutors
- Tutoring was predominantly didactic with extensive knowledge sharing
- Significant proportion involved off-task chatting
- High school sessions had double the message count of primary/middle school sessions
- Tutors contributed over 63% of communications across all levels
Implications:
- Need for more interactive teaching styles, especially for younger students
- Importance of supportive learning environment to enhance participation
- Value of balancing didactic instruction with student engagement
- Need for differentiated tutoring approaches based on grade level
Limitations:
- Data from single country and platform
- Imbalanced distribution across school stages
- Coding scheme may not capture all behaviors
- AI annotation model could have biases
- Analysis limited to two consecutive dialogic behaviors
Future Directions:
- Diversify data sources across countries/platforms
- Refine coding schemes
- Improve reliability of annotation models
- Analyze longer sequences of dialogic behaviors
Title and Authors: "Investigating dialogic interaction in K12 online one-on-one mathematics tutoring using AI and sequence mining techniques" by Deliang Wang, Dapeng Shan, Ran Ju, Ben Kao, Chenwei Zhang, and Gaowei Chen
Published On: November 25, 2024
Published By: Education and Information Technologies