Hidden Infrastructure of Academic Feedback

System key

• Blue: Cognitive
• Red: Instructional
• Green: Behavioral
• Navy: Structural
• Red dashed: Temporal
• Purple: Mechanical
• Orange: Strengths

What feedback is for

Hattie & Timperley, 2007

• Feedback questions
• Where am I going? (Feed Up)
• How am I going? (Feed Back)
• Where to next? (Feed Forward)
• Validity gap
• Feedback levels
• Task (about the work)
• Process (about the strategies)
• Self-regulation (about monitoring)
• Self (about the learner)
• Most feedback answers the first question. Learning requires all three.
• High scores do not mean deep learning. Low scores do not mean poor teaching.

Why students don't read it

• Cognitive reasons
• Feedback arrives after the grade (grade closes the loop; Butler 1988)
• Feedback arrives after the course
• Volume overwhelms processing
• No revision opportunity makes feedback moot
• Behavioral reasons
• Emotional response to grade blocks reading
• No incentive if grade is final
• Don't know how to act on it
• The grade is the full stop. Everything after it is invisible.

The timing problem

Feedback impact decays over time.

• Diagram: chart of feedback impact (vertical, low to high) against time elapsed since submission (Immediate, 1 week, 2 weeks, End of semester); impact falls steeply over time.
• Immediate: High impact, high context
• 1 week: Moderate impact
• 2 weeks: Low impact, context lost
• End of semester: ≈0 impact
• Timely feedback turns performance into learning. Delayed feedback turns it into wallpaper.

Written vs. verbal feedback

• Written
• Permanent, referenceable
• Often dense
• Frequently unread
• Verbal (office hours, in-class)
• Immediate, dialogic
• Allows clarification
• Higher retention
• Hybrid (audio/video feedback)
• Higher engagement
• Personal
• Efficient
• Research: verbal and audio feedback often improve revision quality more than written comments alone.
• The most effective feedback is a conversation, not a document.

The specificity system

From vague to actionable.

• Diagram: a scale from less useful to more useful: Vague, then General, then Specific, then Actionable.
• Vague: "Good job." "Needs work."
• General: "Thesis is weak."
• Specific: "Thesis doesn't yet take a position; it describes but doesn't argue."
• Actionable: "Thesis should answer: what do you want me to understand that I wouldn't already know?"
• Specificity is not about quantity of comments. It is about precision of direction.

Feedback banks

Efficiency through systems.

• Diagram: Concept leads to Much feedback repeats across submissions, and to the feedback bank structure, which leads to time savings and quality benefit.
• Feedback bank structure
• Categorized by error type
• Pre-written comments
• Numbered for copy-paste
• Time savings
• 30 min to build
• 90 min / batch
• vs. 300 min
• Quality benefit
• More consistent
• Less fatigued
• Higher specificity
• Build it once. Use it all semester. Your feedback improves as you get less tired.

How institutions misuse evaluations

• Used for
• Tenure decisions
• Salary
• Contract renewal
• Teaching awards
• Designed for
• Course improvement
• Student feedback
• Misuse consequences
• Grade inflation pressure
• Risk-averse pedagogy
• Chilling effect on rigor
• The instrument was designed for one purpose. It is used for another. The mismatch has consequences.

The identity penalty system

Cumulative disadvantage.

• Women faculty: lower on authority, knowledge, clarity
• Faculty of color: lower on relatability, communication
• Accent-marked faculty: lower on professionalism
• Junior faculty: lower across dimensions
• Diagram: a two-way arrow links these groups to the combined effect.
• Combined effect: Evaluation systems structurally disadvantage those already disadvantaged.
• Bias compounds. A single evaluation is noise. A career of them is a system.

What evaluations are useful for

• Useful
• Identifying extreme outliers
• Tracking trends over time for same instructor
• Student experience data
• Open-ended comments for specific improvement
• Not useful
• Comparing across instructors
• Tenure / promotion decisions
• Measuring learning
• Ranking teaching quality
• Knowing what the tool cannot do is part of knowing how to use it.

How to contextualize your evaluations

• Check N (low N = high variance)
• Check demographics (who responded?)
• Separate quantitative scores from qualitative comments
• Compare to yourself over time, not to others
• Weight comments by specificity
• Ignore single-score outliers
• Statistical literacy is self-defense.

What good teaching feedback looks like

• Mid-semester student feedback (formative, actionable)
• Peer observation (structured, criteria-based)
• Teaching portfolio (evidence-based narrative)
• Student learning outcomes (actual measures of learning)
• Self-reflection protocol
• Teaching quality is too complex for a single instrument. Multiple measures give a more honest picture.

Toward better evaluation systems

• Research-aligned alternatives
• COPUS classroom observation
• SoTL evidence portfolios
• Learning-centered rubrics
• Student artifact analysis
• Institutional changes needed
• Decouple evals from high-stakes decisions
• Mandatory context statements
• Training for evaluators
• Better systems exist. Implementing them requires institutional will.

The evaluation timing matrix

Impact varies by time and work type.

• Diagram: a grid of learning impact (vertical, low to high) against time elapsed since submission (Immediate (Days) to Delayed (Weeks or longer)).
• Immediate draft feedback: High impact, High context
• Delayed draft feedback: Moderate impact, Context fades
• Immediate final feedback: Low-moderate impact, High emotional charge
• Delayed final feedback: Near zero impact, No learning opportunity
• The best feedback window is before the grade, not after it.

The feedback ecosystem

Many inputs. One goal: learning.

• Diagram: Student learning at the center, with two-way arrows to Written comments, Verbal office hours, Peer feedback, Rubric transparency, In-class worked examples, Self-assessment (Metacognition) and Audio/video feedback.
• Disconnected: Student satisfaction scores
• Current practice: over-reliance on written end-of-work comments.
• Recommended: a balanced feedback ecosystem.
• No single feedback mode reaches all students. Design the full ecosystem.

The feedback bank blueprint

A four-category structure.

• Conceptual issues (thesis, argument, evidence): 5–10 pre-written comments; numbered; paste + personalize
• Structural issues (organization, transitions, coherence): 5–10 pre-written comments; numbered; paste + personalize
• Mechanical issues (citation, formatting, sentence-level): 5–10 pre-written comments; numbered; paste + personalize
• Strengths (what to preserve and amplify): 5–10 pre-written comments; numbered; paste + personalize
• Thirty minutes of setup can save hours per batch.

The feedback equation

Better feedback is designed, not accidental.

• Timing + Specificity + Actionability + Revision opportunity = Feedback that changes learning
• When feedback arrives after the grade, after the course, or without a path to revision, it is not feedback. It is record-keeping.