What Evidence Should a Composition Tutor Keep to Prove Progress Is Real?

Progress in composition is easy to claim and harder to prove. A stronger system keeps enough comparable evidence to answer five questions: where did the student start, what changed, what still breaks, does the improvement survive a new task, and is the student becoming more independent?

Direct answer: the evidence trail should include baseline → comparable writing samples → recurring-error patterns → feedback and repair → delayed transfer → timed stability → independence. Scores can be included, but they should not be the only evidence.

This Page Owns the Evidence Architecture

There is already a parent-facing guide on how to track real progress in composition tuition.

This page asks a more technical reader-safe question: what evidence should exist underneath a claim of progress?

Why One Score Is Not Enough

A single composition mark can move because of topic familiarity, school marking differences, adult help, memorised language, time available or simple day-to-day variation.

That does not make scores useless. It means a score should sit inside a wider evidence trail.

Progress is more convincing when several independent signals move in the same direction.

Evidence Layer 1: A Truthful Baseline

The baseline should show what the child can currently do without unusual support.

• Can the child interpret the topic?
• Can the child generate a workable route?
• Can paragraphs remain relevant?
• Does sentence control remain stable?
• Which grammar errors repeat?
• Can the child self-edit?
• How much prompting is needed?

A heavily coached first sample is a poor baseline because later independence cannot be compared fairly.

Evidence Layer 2: Comparable Writing Samples

To judge change, compare work that is similar enough to make the comparison meaningful.

Weak comparison	Stronger comparison
Easy familiar topic vs difficult unfamiliar topic	Two similarly demanding unseen topics
Untimed draft vs timed exam task	Untimed vs untimed, then timed vs timed
Heavily guided piece vs independent homework	Compare support level explicitly
One unusually high score	Pattern across several samples

Perfect experimental control is unnecessary. The purpose is simply to avoid reading too much into incomparable conditions.

Evidence Layer 3: Recurring-Error Reduction

One of the strongest signs of real progress is that a previously recurring weakness appears less often or requires less support.

• tense drift falls across several compositions,
• paragraphs stop wandering away from the topic,
• endings become less abrupt,
• sentence fragments reduce,
• or the child stops forcing memorised phrases into unsuitable contexts.

Tracking the recurring pattern is often more informative than counting every isolated mistake.

Evidence Layer 4: Feedback Followed by Repair

Marked work shows what the tutor noticed. Repair shows what the learner could do with that information.

A useful evidence chain is:

original attempt → tutor identifies issue → student rewrites → corrected version → later unseen retest

If the tutor rewrites the sentence, the page may improve without proving that the student improved.

Evidence Layer 5: Delayed Retrieval

Immediate improvement can depend on fresh memory of the lesson. Retest later.

• Can the student still use the planning principle next week?
• Does the grammar correction survive after several days?
• Does the student remember the self-editing routine without seeing the checklist?

Delayed retrieval separates short-term supported success from more durable learning.

Evidence Layer 6: Transfer to a Changed Task

A student may perform well because the new task resembles the teaching example. Transfer requires some change.

• different topic,
• different picture,
• different narrative problem,
• different paragraph purpose,
• or less scaffolding.

The principle should survive the surface change.

Evidence Layer 7: Timed Stability

Composition quality under no time pressure and composition quality under examination conditions are different measurements.

Track both.

Pattern	Interpretation
Weak untimed and weak timed	Underlying writing capability still needs repair
Strong untimed, weak timed	Execution or automaticity may be the remaining bottleneck
Strong timed only on familiar topics	Transfer may still be weak
Stable across unfamiliar timed tasks	Stronger evidence of exam-ready control

Timing should not be used to conceal a foundation problem. It is a later stress test.

Evidence Layer 8: Independence

Independence is one of the most important progress measures because it reveals whether the learner is carrying more of the writing process.

• fewer tutor prompts,
• less need for model phrases,
• more independent planning,
• better self-detection,
• more accurate questions when stuck,
• and successful work outside tuition.

If marks rise while dependence also rises, the evidence is mixed.

Evidence Layer 9: School Transfer

Tuition should eventually change writing beyond the tuition room.

Useful school-side signals include:

• similar errors reduce in school compositions,
• teacher comments improve in the same area,
• the child handles unfamiliar prompts more calmly,
• or school marks become more stable rather than occasionally spiking.

School evidence should not be expected to change instantly, but it provides an important external receiver for the learning.

What the Evidence Record Does Not Need to Become

Good tracking does not require turning every child into a spreadsheet.

• Do not measure everything because it can be measured.
• Do not create frequent high-stakes testing solely to generate data.
• Do not publish private student records as marketing proof.
• Do not reduce a complex writer to one composite number.

The evidence architecture should support teaching decisions, not create administrative theatre.

A Minimal Progress Evidence Set

Evidence	Question answered
Baseline sample	Where did the child start?
Later comparable sample	What visibly changed?
Recurring-error notes	What is reducing or persisting?
Repair sample	Can the child use feedback?
Delayed unseen task	Did the learning transfer?
Prompt/support level	Is independence increasing?
Timed sample when appropriate	Does capability survive execution pressure?

A Strong Progress Statement

Instead of:

“Your child improved a lot.”

A stronger summary would be:

“The child now plans independently and paragraph relevance is stable across three unseen topics. Tense control still falls under timed writing, so the next target is maintaining accuracy while execution speed increases.”

That statement shows evidence, residual weakness and next action.

At eduKate Punggol

In a maximum three-student, 1.5-hour tutorial, the small group creates enough visibility to compare attempts, identify recurring errors, watch support levels and test whether corrections transfer. The useful record is not the amount of work completed. It is the evidence of movement from Catch Up or Keep Up toward more stable, independent performance.

Continue

• How to Track Real Progress in Composition Tuition
• How Feedback Should Work in Primary Composition Tuition
• How Soon Should Composition Tuition Show Progress?

Evidence principle: real progress should leave a trace that can survive comparison, delay, variation and reduced support.