**Measuring one thing to get information about some other, unmeasurable thing.** Good proxies have a high level of [[correlation]] to the thing they are being used to measure-by-proxy. Bad proxies have a low [[correlation]] to the thing they are a proxy for. They can also lead to [[Rule Beating]]. Proxies are subject to confounding variables. Proxy measures can be either [[Leading Measures]] or [[Lagging Measures]]. # Example Proxy Measures - Tylenol consumption as a proxy for overall pain in your life. - Confounding variable: you share your bottle of Tylenol without others, who will introduce [[Noise]] into the signal. - Infant mortality rate as a proxy for a countries level of health care - GDP as a proxy for the average standard of living in a country - Proxy gap: assumes a relatively even distribution of wealth. If the entire GDP is earned by one person and everyone else in the country makes nothing. - Dates or arguments as proxy measures for how well a relationship is going **** ## Source ## Related - [[KPIs]] - [[Good Metrics]] - [[Day 1]] - [[The Scientific Method]] - [[Metrics]]