Estimation question

How many Pianists are there in New York city?

Practice this question out loud. An AI interviewer asks it, follows up like a real interviewer would, and scores your answer. Type or speak.

What this question tests

A niche market-sizing estimation testing the ability to decompose an unusual, low-base-rate population using layered assumptions.

Clarify scope: assume 'pianists' means people who play piano with meaningful skill or regularity (not anyone who ever touched a piano), including students, hobbyists, and professionals, and state this definition.
Estimate NYC population: roughly 8.5 million people.
Estimate the % who play piano at a meaningful level: assume based on general music participation rates that roughly 2-3% of the population plays piano regularly, a stated assumption drawing on typical instrument-participation studies.
Compute: 8.5M x 2.5% (midpoint) = roughly 210,000 pianists.
Cross-check via a second method: estimate via piano teachers and schools, e.g., assume a few thousand piano teachers in NYC each with 20-30 active students, producing a lower-bound figure in the tens of thousands for actively-taught pianists, then reconcile that this undercounts self-taught/casual players captured in the population-based method.
Present a final range acknowledging the two methods bracket different definitions: state roughly 150,000-250,000 as a reasoned estimate.

Defines 'pianist' explicitly before estimating, since the answer changes enormously depending on whether casual dabblers count.
Uses two independent estimation methods (population participation rate vs teacher/student supply) and reconciles why they differ.
States the participation-rate assumption clearly as a number pulled from general knowledge of music participation, not fabricated as a precise fact.
Delivers a final range rather than one falsely precise number, showing calibrated confidence.

Not defining 'pianist' at all, producing wildly different answers depending on hidden assumptions.
Using only one estimation method with no cross-check.
Presenting a single overly precise number (e.g., '187,432') instead of a reasoned range.

How would your estimate change if we only counted professional/gigging pianists?
Which of your two methods do you trust more, and why?
How would you actually go validate this number if you had a week?