119i - Peer Review of Statistical Reports

Lesson 119 — Peer Review and Refining Statistical Reports

Strand: Statistics | Descriptor: AC9M7ST03 | Duration: 45 minutes

Block note. Review stage. Students review two peers’ reports from Lesson 118 and revise their own for submission in Lesson 120.

Learning Intentions

• To review a statistical report against agreed criteria.
• To give and act on constructive feedback.

Success Criteria

I can:

1. Review a report systematically against a checklist.
2. Identify unsupported claims and missing information.
3. Give feedback that is specific and usable.
4. Revise my own report in response to feedback.

Warmup

(6 minutes — three flawed claims, pairs)

Each claim comes from a real-looking report. Name the flaw and rewrite it.

1. “The average time was .”
2. “Our results prove that Year 7 students prefer maths to English.”
3. “Most students were around the middle.”

Answers: 1. Which average? Units? Sample size? Rewrite: “The median time was seconds across our sample of students.”; 2. “Prove” is too strong, and a class sample cannot support a claim about all Year 7 students. Rewrite: “In our sample, more students chose maths than English.”; 3. Vague — no measure, no numbers. Rewrite: “The median was , with most values between and .”

The three faults to watch for all lesson: unnamed measures, over-claiming, and vagueness.

Activities

Activity 1 — Explicit Instruction: how to Review (10 min)

Reviewing is not fault-finding. The purpose is to make the report usable by a reader — the same standard as Lesson 76’s modelling reports.

The review protocol — three passes:

Pass	Question	Focus
1. Reader’s pass	Could I repeat this investigation and understand the findings?	Clarity and completeness
2. Checker’s pass	Do the numbers agree with the display and with each other?	Accuracy
3. Sceptic’s pass	Does any claim go beyond what the data shows?	Honesty

Feedback that is usable — the difference, modelled:

Unusable	Usable
”Needs more detail."	"Section 1 doesn’t say how you measured arm span — from where to where?"
"Good work!"	"Your limitations paragraph names the small sample and the time-of-day issue — that’s the most honest section here."
"Wrong."	"Your median of doesn’t match your stem-and-leaf plot — I count the middle value as .”

The rule: every comment points at a specific place in the report and says what a reader cannot tell or what does not add up.

The feedback format — two stars and a question, as in Lessons 76 and 106:

• Star 1: something the report does well, named specifically.
• Star 2: a second specific strength.
• Question: one thing a reader still cannot tell, phrased as a question.

Activity 2 — Review Two Reports (20 min)

Rotate reports so each is reviewed twice. Ten minutes per report.

The review sheet — completed for each report reviewed:

Criterion	✓ / ✗ / ?	Comment
Statistical question stated precisely
Method repeatable by a stranger
Variable, units and precision given
Sample size and population stated
Display has title, labels/key, all intervals
Display matches the data
All four measures given
Chosen measure named and justified
Shape, centre, spread and outliers all described
Question answered directly
Prediction compared with result
At least two specific limitations
No claim exceeds what the data supports

Then write: two stars and one question.

Circulating prompts for reviewers:

Prompt	Purpose
Check their median against their own display — do you agree?	Pass 2; the commonest arithmetic slip.
Read only their conclusion. Does it match their evidence?	Pass 3, isolated from the persuasive surrounding text.
Could you collect this data tomorrow from their method alone?	Pass 1’s real test.
Is their limitation specific, or could it apply to any study?	”Small sample” alone is boilerplate; “our 18 students were all from one class, taken after PE” is specific.
Which of their sentences claims the most? Is it earned?	Locates over-claiming precisely.

A note on the checker’s pass: reviewers should actually recount one statistic from the display, not simply tick. Roughly a third of reports contain a median or count error that only recalculation catches — the same lesson as Lesson 105’s algorithm testing.

Activity 3 — Revise (9 min)

Individually. Reports return to their authors with two review sheets.

1. Read both reviews. Where do they agree? That is your priority.
2. Answer each reviewer’s question — in writing, in the report itself.
3. Make at least three specific revisions.
4. Complete the revision log below.

The revision log:

Feedback received	What I changed	Why

The disagreement case, worth naming: if the two reviewers conflict, the author decides — and records the reasoning in the log. Reviewers are not always right, and defending a choice with a reason is as valuable as changing it.

Closing prompt for the class: which criterion was most often marked ✗ across the room? (Expect: limitations too vague, or spread omitted. Whichever it is, name it on the board as the class’s collective weak point.)

Checks for Understanding

(5 minutes — exit ticket, collected with the revision log)

1. Name the three passes of the review protocol and what each looks for.
2. Rewrite as usable feedback: “Your results section isn’t very good.”
3. What makes a limitation specific rather than boilerplate?
4. Name one revision you made and why.
5. Reasoning. Your two reviewers disagreed about a point. Explain how you decided.

Answers: 1. Reader’s (clarity/completeness), checker’s (accuracy), sceptic’s (honesty); 2. E.g. “Your results section gives the median but not the range — a reader can’t tell how spread out the data was.”; 3. It names the actual circumstance of this study (who, when, how) rather than a phrase that would fit any investigation; 4–5. Student’s own, marked on the quality of the reasoning.

Common Misconceptions

Misconception	How to pre-empt it
Reviewing as fault-hunting.	The two-stars-and-a-question format; the purpose is usability.
Vague feedback.	The usable/unusable table; every comment must point somewhere.
Ticking without checking.	The checker’s pass requires recalculating one statistic.
Accepting all feedback uncritically.	The disagreement case — authors decide and justify.
Boilerplate limitations.	The specificity prompt during circulation.
Revising without recording why.	The revision log is a required deliverable.

Enrichment — Competition-Style Problems

E1 (Kangaroo style). A report’s stem-and-leaf plot shows values but the text says “we measured students”. Which is more likely wrong, and how would you check?

Answer

Either could be — check against the raw data sheet, which is the authority. Most often a leaf was omitted when transcribing.

E2 (AMC Junior style). A report gives median , mean , and describes the distribution as “symmetric”. Identify the inconsistency.

Answer

A mean well above the median indicates a right-skew, not symmetry. Either the description is wrong or one of the calculations is — both need rechecking against the display.

E3 (Challenge). A reviewer writes: “Your conclusion says students who read more score higher, but your data only shows they tend to. Is that a fair comment?”

Answer

Yes, and it is precisely the right comment. Data showing two variables moving together supports an association; claiming one produces the other requires ruling out alternative explanations, which a small observational study cannot do.

E4 (Challenge). Why is it valuable to have a report reviewed by someone who did not collect the data?

Answer

The collectors know what they meant, so they fill gaps unconsciously when reading — exactly the sandwich problem from Lesson 103. An outsider can only use what is written, which is the real test of whether the report communicates.

Homework

1. Complete your revision log with all three revisions.
2. Produce the final version of your report.
3. Rewrite each as usable feedback: (a) “Not enough detail.” (b) “Good graph.” (c) “This bit is wrong.”
4. A report states: median , mean , range , and describes the shape as “strongly right-skewed”. Identify the inconsistency and explain.
5. Write a specific limitations sentence for a study of students measured during a wet-weather lunchtime indoors.
6. Name the three review passes and give one question you would ask in each.
7. For your own final report, name the section you are least confident about and say why.
8. Reasoning. Explain why “small sample size” alone is a weak limitation.
9. Reasoning. Explain why a reviewer who did not collect the data is more useful than one who did.
10. Challenge. Write a short review (two stars and a question) of the warmup’s Report B from Lesson 118.

Answers: Q3 — e.g. (a) “Your method doesn’t say what precision you rounded to.” (b) “Your axis labels and key make the display easy to read without explanation.” (c) “Your median doesn’t match your plot — I count , not .” Q4 — a mean essentially equal to the median indicates a roughly symmetric distribution, and a range of only leaves little room for a long tail; the shape description contradicts both figures. Q5 — e.g. “All students were measured indoors during a wet lunchtime, when many had been sitting still for an extended period; reaction times measured after physical activity might differ.” Q8 — it applies to almost every classroom study and tells the reader nothing specific about this one; a useful limitation names how the sample was obtained and why that matters. Q10 — e.g. stars: the method is precise enough to repeat, and the limitation about the sample being one class is clearly stated; question: how were the ruler-drop distances converted to milliseconds?