1 of 45

1

Using Anticipatory Diagrammatic Self-explanation to Support Learning and Performance in Early Algebra

Tomohiro Nagashima1, Anna N. Bartel2, Gautam Yadav1, Stephanie Tseng1,

Nicholas A. Vest2, Elena M. Silla2, Martha W. Alibali2, & Vincent Aleven1

​

1Carnegie Mellon University , 2University of Wisconsin - Madison

tnagashi@cs.cmu.edu

​

​

tomonagashima

2 of 45

2

Background

Self-explanation is an established effective learning strategy

(Bisra et al., 2018; Chi et al., 1989; Rittle-Johnson et al., 2017)

​

From Understanding How We Learn; A Visual Guide By Yana Weinstein and Megan Sumeracki, with Illustrations by Oliver Caviglioli, taken from https://www.learningscientists.org/blog/2020/2/20-1

3 of 45

3

Background

Self-explanation is an established effective learning strategy

(Bisra et al., 2018; Chi et al., 1989; Rittle-Johnson et al., 2017)

  • Through making connections between learner’s prior knowledge and the content they are learning

​

From Understanding How We Learn; A Visual Guide By Yana Weinstein and Megan Sumeracki, with Illustrations by Oliver Caviglioli, taken from https://www.learningscientists.org/blog/2020/2/20-1

4 of 45

4

Background

Self-explanation is an established effective learning strategy

(Bisra et al., 2018; Chi et al., 1989; Rittle-Johnson et al., 2017)

  • Through making connections between learner’s prior knowledge and the content they are learning

​

From Understanding How We Learn; A Visual Guide By Yana Weinstein and Megan Sumeracki, with Illustrations by Oliver Caviglioli, taken from https://www.learningscientists.org/blog/2020/2/20-1

Designing effective self-explanation is hard!

5 of 45

5

Background

Scaffolding self-explanation is a challenging design problem (Bisra et al., 2018)

​

6 of 45

6

Background

Scaffolding self-explanation is a challenging design problem (Bisra et al., 2018)

How to support both learning and problem-solving performance (e.g., error rate, time spent)?

From Long, Y., & Aleven, V. (2017). Enhancing learning outcomes through self-regulated learning support with an open learner model. User Modeling and User-Adapted Interaction, 27(1), 55-88.

7 of 45

7

Background

Scaffolding self-explanation is a challenging design problem (Bisra et al., 2018)

How to support both learning and problem-solving performance (e.g., error rate, time spent)?

From Long, Y., & Aleven, V. (2017). Enhancing learning outcomes through self-regulated learning support with an open learner model. User Modeling and User-Adapted Interaction, 27(1), 55-88.

8 of 45

8

Design: Intelligent Tutoring System (ITS) for Middle-school Algebra

Anticipatory Diagrammatic Self-explanation: learners “explain” their future problem-solving steps

9 of 45

9

Design: Intelligent Tutoring System (ITS) for Middle-school Algebra

Anticipatory Diagrammatic Self-explanation: learners “explain” their future problem-solving steps

10 of 45

10

Design: Intelligent Tutoring System (ITS) for Middle-school Algebra

Anticipatory Diagrammatic Self-explanation: learners “explain” their future problem-solving steps

11 of 45

11

Design: Intelligent Tutoring System (ITS) for Middle-school Algebra

Anticipatory Diagrammatic Self-explanation: learners “explain” their future problem-solving steps

12 of 45

12

Design: Intelligent Tutoring System (ITS) for Middle-school Algebra

Anticipatory Diagrammatic Self-explanation: learners “explain” their future problem-solving steps

13 of 45

13

From Long, Y., & Aleven, V. (2017). Enhancing learning outcomes through self-regulated learning support with an open learner model. User Modeling and User-Adapted Interaction, 27(1), 55-88.

14 of 45

14

Design: Designing for both Learning and Performance

15 of 45

15

Design: Designing for both Learning and Performance

  1. Visual Representations

​

​

​

​

​

​

to support conceptual understanding and strategic decision-making

(Booth & Koedinger, 2012; Rau, 2017)

16 of 45

16

Design: Designing for both Learning and Performance

  • Visual Representations

​

​

​

​

​

​

to support conceptual understanding and strategic decision-making

(Booth & Koedinger, 2012; Rau, 2017)

Tape diagrams

(Bartel et al., 2021; Murata, 2008;

Nagashima, Yang, et al. 2010)

17 of 45

17

Design: Designing for both Learning and Performance

  • Visual Representations

​

​

​

​

​

​

to support conceptual understanding and strategic decision-making

(Booth & Koedinger, 2012; Rau, 2017)

Tape diagrams

(Bartel et al., 2021; Murata, 2008;

Nagashima, Yang, et al. 2010)

2. Contrasting Cases

​

​

​

to support understanding of correct and strategic problem-solving steps through explicit comparisons

(Schwartz et al., 2011)

18 of 45

18

Design: Designing for both Learning and Performance

  • Visual Representations

​

​

​

​

​

​

to support conceptual understanding and strategic decision-making

(Booth & Koedinger, 2012; Rau, 2017)

Tape diagrams

(Bartel et al., 2021; Murata, 2008;

Nagashima, Yang, et al. 2010)

2. Contrasting Cases

​

​

​

to support understanding of correct and strategic problem-solving steps through explicit comparisons

(Schwartz et al., 2011)

19 of 45

19

Design: Designing for both Learning and Performance

  • Visual Representations

​

​

​

​

​

​

to support conceptual understanding and strategic decision-making

(Booth & Koedinger, 2012; Rau, 2017)

Tape diagrams

(Bartel et al., 2021; Murata, 2008;

Nagashima, Yang, et al. 2010)

2. Contrasting Cases

​

​

​

to support understanding of correct and strategic problem-solving steps through explicit comparisons

(Schwartz et al., 2011)

3. Anticipative Reasoning

​

​

​

​

​

​

​

​

​

​

​

​

​

​

to support inference generation about strategic problem-solving steps, which would lead to efficient and effective learning

(Nagashima et al., 2021; Renkl, 1997)

What would be a good next step…?

20 of 45

20

Design: Designing for both Learning and Performance

  • Visual Representations

​

​

​

​

​

​

​

​

​

​

​

​

​

​

2. Contrasting Cases

​

​

​

​

​

​

​

​

​

​

​

​

​

​

3. Anticipative Reasoning

​

​

​

​

​

​

​

​

​

​

​

​

​

​

to support conceptual understanding and strategic decision-making

to support understanding of correct and strategic problem-solving steps through explicit comparisons

(Booth & Koedinger, 2012; Rau, 2017)

to support inference generation about strategic problem-solving steps, which would lead to efficient and effective learning

(Renkl, 1997)

What would be a good next step…?

(Schwartz et al., 2011)

Tape diagrams (Bartel et al., 2021; Murata, 2008; Nagashima, Yang, et al. 2010)

21 of 45

21

Method: ”In-vivo” Classroom Research

  • 108 students with 4 teachers, across 9 classes at 2 schools in the US
  • Pre/Posttests to measure conceptual understanding and procedural skills

22 of 45

22

Method: ”In-vivo” Classroom Research

  • 108 students with 4 teachers, across 9 classes at 2 schools in the US
  • Pre/Posttests to measure conceptual understanding and procedural skills

23 of 45

23

Method: ”In-vivo” Classroom Research

  • 108 students with 4 teachers, across 9 classes at 2 schools in the US
  • Pre/Posttests to measure conceptual understanding and procedural skills

24 of 45

24

Method: ”In-vivo” Classroom Research

  • 108 students with 4 teachers, across 9 classes at 2 schools in the US
  • Pre/Posttests to measure conceptual understanding and procedural skills

25 of 45

25

Method: ”In-vivo” Classroom Research

No-Diagram condition: Tutor with no anticipatory diagrammatic self-explanation

26 of 45

Results: Learning

Did students learn from pretest to posttest?

27 of 45

*

No significant gains between conceptual or procedural knowledge pretest/posttest as a function of condition

No significant overall gain in procedural knowledge from pretest to posttest

Significant overall gain in conceptual knowledge from pretest to posttest

Results (Learning):

Did students learn from pretest to posttest?

28 of 45

28

Results (Learning):

What strategies did students use when solving problems?

29 of 45

29

Results (Learning):

What strategies did students use when solving problems?

Strategy name

Description

Algebra

Student uses algebraic manipulations to find an answer

Unwind

Student works backward using inverse operations to find an answer

Guess and Check

Student tests potential solutions by substituting different values

Other

Student uses other non-algebraic strategies

Answer Only

Student provides an answer without showing any written work

No Attempt

Student leaves problem blank or explicitly indicates that she/he does not know how to solve the problem

From Chu et al. (2017) & Koedinger et al. (2008)

30 of 45

30

Strategy name

Description

Algebra

Student uses algebraic manipulations to find an answer

Unwind

Student works backward using inverse operations to find an answer

Guess and Check

Student tests potential solutions by substituting different values

Other

Student uses other non-algebraic strategies

Answer Only

Student provides an answer without showing any written work

No Attempt

Student leaves problem blank or explicitly indicates that she/he does not know how to solve the problem

From Chu et al. (2017) & Koedinger et al. (2008)

Non-Algebra

strategies

Algebra

strategy

Results (Learning):

What strategies did students use when solving problems?

31 of 45

Students with anticipatory diagrammatic self-explanation made a transition to using a formal algebraic problem-solving strategy

​

(McNemar’s test; p < .01)

No Diagram Condition

Diagram Condition

Results (Learning):

What strategies did students use when solving problems?

32 of 45

Results: Performance

Did students benefit from the intervention during the learning task?

33 of 45

33

Results: Performance within the ITS

Students with anticipatory diagrammatic self-explanation showed lower error rate on symbolic steps (only).

34 of 45

34

Results: Performance within the ITS

Students with anticipatory diagrammatic self-explanation showed lower error rate on symbolic steps (only).

Diagram

35 of 45

35

Results: Performance within the ITS

Students with anticipatory diagrammatic self-explanation showed lower error rate on symbolic steps (only).

Diagram

No-Diagram

36 of 45

36

Results: Performance within the ITS

Students with anticipatory diagrammatic self-explanation showed lower error rate on symbolic steps (only).

Diagram

No-Diagram

37 of 45

How did students perform while problem-solving in the ITS? 

Students with anticipatory diagrammatic self-explanation showed efficient learning:

Less time per step

Trend toward fewer hints

No difference problems solved

p = .03

p = .07

p = .60

38 of 45

38

Discussion

Anticipatory diagrammatic self-explanation embedded in an ITS supported

both learning and performance:

​

39 of 45

39

Discussion

Anticipatory diagrammatic self-explanation embedded in an ITS supported

both learning and performance:

  • Learning: Students acquired a formal algebraic strategy

​

40 of 45

40

Discussion

Anticipatory diagrammatic self-explanation embedded in an ITS supported

both learning and performance:

  • Learning: Students acquired a formal algebraic strategy
  • Performance: Students solved problems more efficiently
    • spent less time per step on symbolic steps
    • used fewer hints on symbolic steps
    • showed a lower error rate on symbolic steps

​

41 of 45

41

Contribution

Scaffolding self-explanation is a challenging design problem

42 of 45

42

Contribution

Scaffolding self-explanation is a challenging design problem

Anticipatory diagrammatic self-explanation provides an example of how visual representations, contrasting cases, and anticipative reasoning can be integrated into effective scaffolding that helps students learn and perform well.

​

43 of 45

43

Acknowledgements

This research was supported by NSF Award #1760922 and by the Institute of Education Sciences, U.S. Department of Education, through Award #R305B150003 to the University of Wisconsin–Madison. We thank Lauren E. Anthony, Max Benson, Susan Brunner, Octav Popescu, Jonathan Sewall, and all participating teachers and students.

​

​

​

44 of 45

44

Thank you!

This research was supported by NSF Award #1760922 and by the Institute of Education Sciences, U.S. Department of Education, through Award #R305B150003 to the University of Wisconsin–Madison. We thank Lauren E. Anthony, Max Benson, Susan Brunner, Octav Popescu, Jonathan Sewall, and all participating teachers and students.

​

​

​

tnagashi@cs.cmu.edu

tomonagashima

45 of 45

45

Thank you!

This research was supported by NSF Award #1760922 and by the Institute of Education Sciences, U.S. Department of Education, through Award #R305B150003 to the University of Wisconsin–Madison. We thank Lauren E. Anthony, Max Benson, Susan Brunner, Octav Popescu, Jonathan Sewall, and all participating teachers and students.

​

​

​

tnagashi@cs.cmu.edu

tomonagashima

(#470) Reasoning about Equations with Tape Diagrams: Insights from Students and Math Teachers

Tomorrow/later today!!

Anna Bartel