Collaborative peer feedback

<!--

-->
            <p style="margin-top: 23vh; text-align: center;">UoN P&A Mathematics Seminar | 2025 Oct 21</p>

# Collaborative peer feedback

### Dave Smith

---

## Acknowledgement

Project supported by 2017 Teaching Innovation Grant from the Centre for Teaching & Learning at Yale-NUS College.

### Publication
            D. A. Smith *Collaborative peer feedback*, Proceedings of ICEduTech 2017, Ed. P. Kommers, T. Issa, P. Isías, A. Hol, IADIS (2017), 183–186.

### Implementations
            * **2017–2024** Proof course at Yale-NUS College, Singapore.
            * **2025–** MATH2350 at University of Newcastle, Australia
            ---

## What is peer feedback?

#### Process
            * Students complete an assignment (summative) or a first draft (formative).
            * Students "mark" each other's work, giving feedback.
            * Students respond to the feedback they received.
            
            ---

## What is peer feedback?

#### Popular models
            * Moore method: presentation to whole class followed by criticism.
            * Peer feedback is homework, through VLE (Canvas).
            * Bring homework to class, exchange with neighbour, read, converse.
            
            ---

## What is peer feedback?

#### Grading (stakes)
            * Peer feedback learning activities need not include any grade component.
            * Course score can be awarded for active participation in learning activity.
            * Some models include student provided grades counting toward course score.
            * In some models, students grade their own work based on the feedback they have received from their peers.
            
            ---

## Why collaborative peer feedback?
            
            ---

## Why <span class="deemph">collaborative peer</span> feedback?
            
            ---

## Why <span class="deemph">collaborative peer</span> feedback?

Because it works very well.

> Feedback has extraordinarily large and consistent positive effect on learning compared to other aspects of teaching.

Black & William *Assessment and classroom learning* (1998)

---

## Why <span class="deemph">collaborative peer</span> feedback?

Because it works very well.

> Feedback has extraordinarily large and consistent positive effect on learning compared to other aspects of teaching.

Black & William *Assessment and classroom learning* (1998)

Because it works better than anything else.

> Feedback is the most important single influence on student achievement.

Hattie *Identifying the salient facets of a model of student learning: a synthesis of meta-analyses* (1987)

---

## Why <span class="deemph">collaborative peer</span> feedback?

Because it works very well.

> Feedback has extraordinarily large and consistent positive effect on learning compared to other aspects of teaching.

Black & William *Assessment and classroom learning* (1998)

Because it works better than anything else.

> Feedback is the most important single influence on student achievement.

Hattie *Identifying the salient facets of a model of student learning: a synthesis of meta-analyses* (1987)

And students know it.

> Mathematics students want specific, detailed, facilitative feedback.

Rice et. al. *Improving student feedback in distance education: a research report* (1994)

---

## Why <span class="deemph">collaborative</span> peer feedback?

---

## Why <span class="deemph">collaborative</span> peer feedback?

Instructor time.

---

## Why <span class="deemph">collaborative</span> peer feedback?

Instructor time.

The provider of feedback can also benefit.

<div class="twocols">
              <div>
                <p>Submitting student must</p>
                <ul>
                  <li>find data</li>
                  <li>construct arguments</li>
                  <li>evaluate methodologies</li>
                </ul>
              </div>
              <div>
                <p>Feedback providor must</p>
                <ul>
                  <li>check data</li>
                  <li>interrogate arguments</li>
                  <li>appraise evaluations</li>
                </ul>
              </div>
            </div>

---

## Why <span class="deemph">collaborative</span> peer feedback?

Instructor time.

The provider of feedback can also benefit.

The feedback provider usually attains deeper learning outcomes.

> Self-, peer and collaborative assessment helps to remove the student / tutor barrier, develop enterprising competencies in students and can lead to greater motivation and ‘deeper’ learning.

Somervell *Issues in Assessment, Enterprise and Higher Education: the case for self-, peer and collaborative assessment* (1993)

---

## Why <span class="deemph">collaborative</span> peer feedback?

Instructor time.

<div>

The provider of feedback can also benefit.

The feedback provider usually attains deeper learning outcomes.

> Self-, peer and collaborative assessment helps to remove the student / tutor barrier, develop enterprising competencies in students and can lead to greater motivation and ‘deeper’ learning.

Somervell *Issues in Assessment, Enterprise and Higher Education: the case for self-, peer and collaborative assessment* (1993)

</div>

---

## Why <span class="deemph">collaborative</span> peer feedback?

Instructor time.

<div>

The provider of feedback can also benefit.

The feedback provider usually attains deeper learning outcomes.

> Self-, peer and collaborative assessment helps to remove the student / tutor barrier, develop enterprising competencies in students and can lead to greater motivation and ‘deeper’ learning.

Somervell *Issues in Assessment, Enterprise and Higher Education: the case for self-, peer and collaborative assessment* (1993)

</div>

---

## Dangers of peer feedback?

* Feedback provider may not engage seriously with task
            * low quality
            * high variance in quality
            * receiving student doubts quality

---

## Dangers of peer feedback?

* Feedback provider may not engage seriously with task
            * low quality
            * high variance in quality
            * receiving student doubts quality
            * anonymity valuable in both roles

> When students perceived that their teacher took care not to embarrass them, they reported lower levels of maladaptive behaviors.

Friedel et. al. *"Stop embarrassing me!" Relations among student perceptions of teachers, classroom goals, and maladaptive behaviors* (2002)

---

## The Proof course at Yale-NUS College

Mostly a course in mathematical writing.

Learning goals:
            * writing proof
            * critical reading of proof
            * (board) presentation of proof
            * typesetting mathematics with LaTeX
            * methods of proof: contradiction, induction, $\varepsilon$ $:$ $\delta$ arguments
            * somef oundational topics: elementary number theory, combinatorics, functions and sets, field arithmetic, construction of $\mathbb N, \mathbb Z, \mathbb Q, \mathbb R, \mathbb C$

Enrollment
            * 40% year 2 students
            * 15—23 students per section; approx 45 students per year
            * 80% went on to major in Mathematics, Computer Science & Statistics
            * Highly motivated students

---

## Why peer feedback in Proof?

Peer feedback risky because
            * submitting student not used to precise writing
            * feedback provider not used to unpicking imprecise writing
            * low probability of valuable feedback, especially early in course

---

## Why peer feedback in Proof?

Potential advantages
            * personalised feedback especially valuable on new mode of communication
            * "low quality" feedback resulting from feedback provider misunderstanding often represents failure of submitting student to write clearly
            * critical reading of proofs of varying quality is a learning goal
            * easier to practivce critical reading of another's proof before one's own
            * sharing LaTeX tips between learners

---

## Why peer feedback in Proof?

Effective (peer or instructor) feedback on written proofs requires markup directly on submitted document.

---

## Collaborative peer feedback

Aim to address: feedback provider may not engage seriously with task, low quality, high variance of quality, receiving student doubts quality, anonymity valuable in both roles.

<div class="twocols">
              <div>
                <h4>Before class</h4>
                <ul>
                  <li>students submit work (LaTeX source) through Canvas VLE</li>
                  <li>instructor pseudonymises and prints</li>
                  <li>instructor assigns students nad submitted work to feedback groups</li>
                </ul>
                <h4>During class</h4>
                <ul>
                  <li>instructor provides model solutions</li>
                  <li>feedback generated <strong>collaboratively</strong> by groups of 3–4 students</li>
                  <li>groups study submissions together & agree each piece of feedback before writing</li>
                  <li>instructor facilitates generic feedback</li>
                </ul>
              </div>
              <div>
                <h4>After class</h4>
                <ul>
                  <li>instructor scans feedback & maps pseudonyms to names</li>
                  <li>instructor distributes feedback through VLE</li>
                  <li>students complete reflective assignment based on feedback they received</li>
                </ul>
                <h4>Grading</h4>
                <ul>
                  <li>on submission, participation and completion of reflection</li>
                  <li>total 20% of course grade for 9 exercises in 13 week semester</li>
                </ul>
              </div>
            </div>

---

## Effectiveness of collaborative peer feedback in Proof

#### Course evaluations
            * ⅔ of students mentioned the peer feedback exercises when asked about "mode if interaction" or "most effective learning activities".
            * Almost all mentions of peer feedback were positive.
            * After number of peer feedback exercises was increased and generic feedback sessions introduced, all comments were positive.

#### Colleagues' observations of classes
            * Students have "surprisingly deep discussions of their peers' proofs".
            * "Students appear to enjoy" the collaborative peer feedback exercise.
            * The collaborative peer feedback exercise "forced student to challenge their assumptions and those of their peers in a supportive and respectful environment".
            * Collaborative peer feedback "(1) requires students to read lots of different proofs for a small set of mathematical statements and (2) provides unbiased ... feedback to each student on how understandable their proofs were to peers".
            ---

## Technology in collaborative peer feedback

Significant instructor time could be spent on
            * instructor pseudonymises and prints
            * instructor assigns students to feedback groups and submitted work to feedback groups
            * instructor provides model solutions
            * instructor scans feedback & maps pseudonyms to names
            * instructor distributes feedback through VLE

---

## Technology in collaborative peer feedback

Significant instructor time could be spent on
            * instructor pseudonymises and prints 
            * instructor assigns students to feedback groups and submitted work to feedback groups 
            * instructor provides model solutions
            * instructor scans feedback & maps pseudonyms to names 
            * instructor distributes feedback through VLE

Automated in VBA (macro enabled Excel workbook).

Automated in Python. Currently only Mac & Linux, but Windows coming soon.

Less work then expected as only one model solution necessary and can be reused each semester.

---

## Technology in collaborative peer feedback

Automated in VBA (macro enabled Excel workbook).

Automated in Python. Currently only Mac & Linux, but Windows coming soon.

Less work then expected as only one model solution necessary and can be reused each semester.

None of this requires students submit in LaTeX; a pdf submission works too.

---

## Demonstration

<div class="threecols">
              <div>
                <h3>Submission</h3>
                <p style="text-align: center;"><img alt="submission" src="gfx/AmonAmarthSubmission.png" width="450px"></p>
              </div>
              <div>
                <h3>Pseudonymisation</h3>
                <p style="text-align: center;"><img alt="pseudonymisation1" src="gfx/AmonAmarth1.png" width="300px"></p>
                <p style="text-align: center;"><img alt="pseudonymisation2" src="gfx/AmonAmarth2.png" width="300px"></p>
              </div>
              <div>
                <h3>Feedback</h3>
                <p style="text-align: center;"><img alt="feedback1" src="gfx/AmonAmarthFeedback1.png" width="300px"></p>
                <p style="text-align: center;"><img alt="feedback2" src="gfx/AmonAmarthFeedback2.png" width="300px"></p>
              </div>
            </div>

---

## Collaborative peer feedback in MATH2350 at UoN

#### Motivation
            * Most students solve very few problems each week.
            * Students get very little individual feedback on their work.
            * Students are expected to produce more sophisticated proofs than in previous courses.
            * Students have not previously critically read proof.

#### Adaptations
            * Approx 35 minutes of each 50 minute tutorial is dedicated to collaborative peer feedback.
            * Students submit pdf through Canvas LMS.
            * More model solutions provided, including incorrect attempts by the instructor.
            * Class discussion of the pack of model solutions before beginning the peer feedback.
            * Students only get feedback from 1 group.
            * Generic feedback session dropped as students were not willing to participate.

#### Student feedback
            * None yet as the semester is ongoing.
            * Students are more actively engaged on collaborative peer feedback than on worksheet.

---

## Materials

<div class="threecols">
              <div>
                <h3>Canvas assignments</h3>
                <p style="text-align: center;"><img alt="submission" src="gfx/2350-canvas.png" width="450px"></p>
              </div>
              <div>
                <h3>Correct solutions</h3>
                <p style="text-align: center;"><img alt="pseudonymisation1" src="gfx/2350-soln-1.png" width="450px"></p>
              </div>
              <div>
                <h3>Incorrect solutions</h3>
                <p style="text-align: center;"><img alt="feedback1" src="gfx/2350-soln-2.png" width="450px"></p>
              </div>
            </div>

---

## Demonstration

<div class="threecols">
              <div>
                <h3>Submission</h3>
                <p style="text-align: center;"><img alt="submission" src="gfx/2350-submission.png" width="450px"></p>
              </div>
              <div>
                <h3>Pseudonymisation</h3>
                <p style="text-align: center;"><img alt="pseudonymisation1" src="gfx/2350-anon.png" width="450px"></p>
              </div>
              <div>
                <h3>Feedback</h3>
                <p style="text-align: center;"><img alt="feedback1" src="gfx/2350-feedback.png" width="450px"></p>
              </div>
            </div>

---

# Thanks!

### These slides

## ![qr code for URL of slides](q/collaborative-peer-feedback-2025-10-21.svg)