Introducing Reality into Data Metrics
summary
Claims that common metrics mislead are illustrated with concrete failures: a 1948 poll called telephone-sampled poll bias that misreports Dewey vs Truman because phone ownership skews the population; the 2008 mortgage models that assume ever‑rising house prices and bundle subprime loans into CDOs that collapse; the 1998 MMR paper based on 12 cherry-picked subjects with conflicts of interest; life‑expectancy gains inflated by ignoring infant mortality; proposes the TAP method (Tools, Analysis, Population) and stresses critical thinking as the corrective.
AI-written summary · about
chapters (10)
- 0:00Introduction and welcome
- 1:03Future of work and AI impact
- 6:25Critical thinking in data science
- 7:091948 Dewey-Truman polling error
- 9:172008 financial crisis analysis
- 12:23Life expectancy data skew
- 14:31Vaccine study controversy
- 17:01Poverty and crime correlation
- 19:08Resilience of human skills
- 22:16Q&A and critical thinking maintenance
related talks
05mPerception and bias and metrics, oh my!
27mThe Subtle Art of Lying with Statistics
37mHow to Start Critical Thinking in Data Science
40mHow to Lie With Stats and Charts
32mThe Subtle Art of Lying with Statistics
52mMaking Better Decisions with Data
43mThe Subtle Science of Misleading with Statistics
26mMetrics-driven Data Architectures