The Wrong Unit of Analysis: On studying the entity that acts, not the one you can count

I was rigorous about the wrong thing. The measurement was clean. The sampling was defensible. The answer was precise, and it was precise about an entity that was not the one doing the acting. That is the error this book is about, and it is more common than any sampling mistake I know.

A unit of analysis is just the entity a study is actually about: the person, the household, the firm, or the supply chain whose behaviour you are trying to explain. Pick it wrong and every later step inherits the mistake. You can infer about individuals from group data and land in the ecological fallacy, or infer about a group from individual data and land in its mirror. Both produce a confident, well-defended answer about the wrong actor.

I am not a methodologist. I run three small businesses in Indonesia, and I learned this the slow way, by paying for analyses that were correct and useless at the same time. The book argues one thing without flinching: a precise result about the wrong unit is worse than no result, because no result keeps you honest while the wrong one sends you to act where the leverage isn't.

If you choose your analytic unit before you measure, if you ask who actually acts before you ask what to count, you will waste less. That is the whole claim.

This book covers the unit of analysis, the levels-of-analysis problem, the ecological and atomistic fallacies, why precision about the wrong unit misleads, and how to match a decision or an intervention to the entity that does the acting. Written from an Indonesian operator's vantage, but the diagnostics apply broadly to other emerging-market and SME settings.