Attendance Memo: Correlation or Cause?
At a district family-engagement meeting, a coordinator presents a six-week text-message pilot. Five of the district's 14 schools sent automated attendance nudges to 1,850 families; the other nine schools continued existing outreach. Average daily attendance in pilot schools rose from 91.4% to 93.1%, while comparison schools rose from 91.6% to 92.0%. The memo concludes, 'The text campaign caused a 1.7-point attendance gain and should be expanded districtwide next semester.' A committee member notes that the five pilot schools also held two new parent-night events during the same period, that participation was chosen by principals rather than assigned, and that the pilot included only six weeks of data. The committee must decide whether to recommend expansion, a larger study, or no change, and must explain its reasoning in a brief public memo.
1. Which recommendation is best supported by the evidence: immediate districtwide expansion, a larger controlled study, or no change? Explain using at least two specific details.
Read the worked answer
Best recommendation: a larger controlled study.
The pilot is promising but does not justify districtwide expansion. The pilot schools gained 1.7 points, but comparison schools gained 0.4 points, so the relative gain is about 1.3 points, not 1.7. More important, the comparison is confounded: principals chose the pilot schools, the pilot schools added two parent nights, and the data cover only six weeks. Those facts support a plausible association between the text campaign and attendance, not a causal claim. 'No change' would ignore a promising signal, but 'expand now' would scale an unverified intervention. A larger study should randomize or match schools, track prior attendance trends, separate the text campaign from parent nights, and run long enough to rule out seasonal effects.
2. The memo says the campaign 'caused' a 1.7-point gain. Rewrite that sentence to match the evidence and explain the change.
Read the worked answer
Revised sentence: 'Over six weeks, pilot schools' average daily attendance rose 1.7 points, compared with 0.4 points in comparison schools; this 1.3-point relative gain suggests the text campaign may be associated with improved attendance, but a larger study is needed before districtwide expansion.'
The revision changes three things. It reports the comparison group, because a 1.7-point rise alone does not show the campaign's effect. It uses 'may be associated' instead of 'caused,' because the pilot was not randomly assigned and another change, the parent nights, occurred at the same time. It states the limit of the six-week window. The original sentence overstates both the size of the effect and the certainty of the causal inference.