The course turns on one question: how wrong can you be, and still be useful? Students learn to build and defend estimates from incomplete evidence, make their assumptions and uncertainty visible, and judge how much precision a decision actually needs. No statistics prerequisite.
What you will do
Alongside a semester-long calibration diary — ranges committed before you know the answer — each week gives you a different way estimates go wrong: decomposition, borrowed evidence, disagreement, pressure, aggregation, stakes, forecasting. You decompose real quantities into chains of assumptions, give each one a stated range, and defend it when someone leans on it. The two pieces of graded work are the diary itself and one real, contested estimate, cross-examined in front of the class.
Who it is for
Anyone who has to give a number to someone who will act on it, without time to run a full study first — the course teaches judgment under incomplete evidence, not formulas.
How the course runs
Each teaching week runs one lecture, delivered by the course convenor, and one tutorial per student, chosen from four weekly timeslots — all four cover the same session content, taught by one of three tutors. There is a two-week teaching break between Weeks 6 and 7, with no lecture or tutorial in either of those weeks. Full times, tutors and the semester overview are on the Sessions page.
Assessment
Two pieces of work, no final examination:
- Calibration Diary — 45%
- Client Estimate & Estimate Court — 55%