Four-Day Week Pilot at ExecuTech
ExecuTech, a 300-person software firm, ran a six-month four-day-week pilot in its Austin office (60 employees). During the pilot, the Austin office's quarterly output rose 9% over the same quarter last year, and sick days fell from 4.1 to 2.7 per employee. The CEO announced that rolling out the four-day week to all 300 employees would raise company-wide output by at least 9%. Unknown to the CEO, the Austin office had installed a new project-management platform two months before the pilot, and the pilot overlapped with the company's slowest season, when output typically dips. The other offices use different products and have different client deadlines. HR measured output as closed tickets, not revenue or quality. No control group was used, and employees volunteered for the pilot. The CEO's bonus is tied to company-wide ticket closure. A board member asks whether the pilot's results justify the rollout.
1. Assess the CEO's inference. Identify at least two rival explanations for the 9% output increase and explain how each weakens the causal claim.
Read the worked answer
The CEO treats a single-office before/after comparison as a causal estimate. At least two rival explanations are visible. (1) The new project-management platform was installed two months before the pilot; if it reduced ticket-handling time, the 9% rise could reflect the platform, not the four-day week. (2) The pilot used volunteers, so the Austin employees may have been more motivated or more productive to begin with; volunteering also creates a Hawthorne effect. In addition, output was measured as closed tickets, which can rise if tickets are split or if quality falls. The comparison to the same quarter last year helps with seasonality, but one office and one year cannot rule out a coincidental good quarter. Because these alternatives predict the same 9% rise, the inference to a company-wide causal effect is underdetermined.
2. What evidence would most weaken the CEO's conclusion?
Read the worked answer
The strongest weakening evidence would be a control comparison: other ExecuTech offices without the four-day week showed similar or larger output increases over the same period, or Austin's increase disappeared after controlling for the platform adoption. Alternatively, revenue or quality data could show that closed tickets rose while actual output fell. Evidence that the volunteers were already the firm's highest performers, or that ticket splitting accounted for the entire 9%, would also undercut the claim. Each item attacks a specific rival explanation rather than merely asserting that the pilot was imperfect.
3. The board asks whether the pilot is a sufficient or necessary condition for the rollout. How should they think about it?
Read the worked answer
The pilot is neither necessary nor sufficient. It is not sufficient because one uncontrolled, volunteered pilot in one office cannot establish that the four-day week caused the gain or that the gain will generalize to 240 other employees with different products and clients. It is not necessary because the company could adopt the policy on other grounds-for example, a randomized trial elsewhere, retention benefits, or employee preferences-without this pilot. The board should ask what decision the pilot is meant to support: if the goal is a causal company-wide productivity forecast, a controlled rollout with matched offices would be more informative.