The sentence that goes furthest
Ranking four true-sounding sentences by how much of each the text actually supports.
Passage
From a report on a school breakfast club:
"Attendance was recorded for the 340 pupils at the school across the autumn term. Pupils who attended the club were absent on 4.1 percent of days, against 6.8 percent for pupils who did not. The club was open to all and pupils chose whether to come. Families were not asked why they did or did not attend."
Four sentences have been drafted for the summary.
Answers in order, explanations withheld, and a report on where the chain broke.
- 42.1Central ideas and detailsStandard
Which sentence states only what the report measured?
- APupils who attended the breakfast club were absent on fewer days than pupils who did not.
- BThe breakfast club reduced absence among pupils who attended.
- CBreakfast clubs improve school attendance.
- DPupils who skip breakfast are more likely to be absent.
Show the answer and reasoningHide the reasoning
Pupils who attended the breakfast club were absent on fewer days than pupils who did not.
The report measured absence rates in two groups and the difference between them. Everything else in the list adds a cause, a generalisation to other schools, or a mechanism, none of which was measured.
- 42.2InferencesStandard
Builds on: Question 1 separated the measurement from the causal claim. This asks what in the report explains why that gap cannot be closed from these figures.
Which detail in the report most weakens a causal reading of the difference?
- AThe school has 340 pupils.
- BAttendance was recorded across one term.
- CPupils chose for themselves whether to attend.
- DThe difference was 2.7 percentage points.
Show the answer and reasoningHide the reasoning
Pupils chose for themselves whether to attend.
Self-selection means the two groups may have differed before the club existed — in household routine, in distance from school, in a dozen things the report did not record. That is what stops the comparison being an experiment.
- 42.3Command of evidenceChallenge
Builds on: Question 2 named self-selection as the problem. The right evidence is whatever that specific problem cannot manufacture, which rules out the first and last choices however large they are.
Which further finding would do most to support a causal reading?
- AA second term of data showing the same 2.7-point gap.
- BAbsence falling among club attenders in the term after the club opened, with no fall among non-attenders.
- CA survey in which attenders say the club helps them arrive on time.
- DA larger school showing a similar gap.
Show the answer and reasoningHide the reasoning
Absence falling among club attenders in the term after the club opened, with no fall among non-attenders.
A change that arrives with the club, in the group exposed to it and not in the group that was not, is the shape of evidence self-selection cannot produce on its own. Repeating the same cross-sectional gap — for another term or another school — repeats the same ambiguity at a larger size.
- 42.4Central ideas and detailsChallenge
Builds on: Question 1 established the floor of what was measured. Ranking the rest by distance from it is what this question asks, and the answer is the one that travels furthest in two directions at once.
Which of the four drafted sentences goes furthest beyond the report?
- APupils who attended were absent on fewer days than pupils who did not.
- BThe club reduced absence among pupils who attended.
- CBreakfast clubs improve school attendance.
- DPupils who skip breakfast are more likely to be absent.
Show the answer and reasoningHide the reasoning
Breakfast clubs improve school attendance.
Two of these add a cause to one school; this one adds a cause and then generalises it to breakfast clubs everywhere, from a single term at a single school of 340. The last choice is also unsupported — nobody was asked about breakfast — but it stays within the school.