The Clinic Waiting-Room Study
A hospital pilot tested a new waiting-room design. In the old design, patients waited an average of 34 minutes and 22% reported dissatisfaction. In the new design, average wait was 31 minutes and 19% reported dissatisfaction. The pilot ran for six weeks in one clinic, during a period when the clinic also hired two additional nurses. Satisfaction surveys were collected only from patients who stayed for a follow-up appointment. The hospital's communications team drafts a press release: "New waiting-room design cuts patient dissatisfaction by 14% and shortens waits by 3 minutes."
1. Assess the press release's two numerical claims against the study as described.
Read the worked answer
First claim: dissatisfaction fell from 22% to 19%, a drop of 3 percentage points. As a relative reduction, 3/22 ≈ 13.6%, which the release rounds to 14%. So the 14% figure is a relative reduction, not an absolute one - readers may misread it as 14 points. The release should say "3 percentage points" or specify "relative." Second claim: waits fell from 34 to 31 minutes, a 3-minute drop - arithmetically correct. But both claims attribute causation to the waiting-room design, and the study cannot support that: two nurses were hired during the same period, and that staffing change plausibly affects both wait time and satisfaction. The release presents correlational before/after data as causal. A defensible version would report the raw before/after numbers and explicitly note the confound.
2. The survey was collected only from patients who stayed for a follow-up. How does this affect the strength of the dissatisfaction finding?
Read the worked answer
It introduces selection bias. Patients who return for follow-up are, by definition, patients who remained engaged with the clinic - plausibly more satisfied or more invested than patients who never returned. Their dissatisfaction reports therefore likely understate the dissatisfaction of the full patient population. The direction of bias matters: the 19% figure is probably a floor, not a representative estimate. Combined with the staffing confound, the finding is doubly weakened. A rigorous design would survey all patients (e.g., at exit) regardless of follow-up status, and would compare periods with stable staffing or randomize the design across clinics.