Life Skills, Episode 7: Reading risk numbers. A 6-minute plain-language lesson: turn every percentage into 'out of 1,000 people', and always ask 'absolute or relative?' (convert units before you compare, like Mbps vs MB/s). Natural frequencies raised correct answers from 16% to 46% (Gigerenzer & Hoffrage 1995); physicians 1/24 to 16/24 (Hoffrage 2000). Worked example with made-up round numbers: 1% prevalence, 90% detection, 9% false alarms -> 9 of 98 positives, about 9%, not 90%. Mammography fact box: ~5 vs ~4 deaths per 1,000 women = 1 fewer in 1,000 (absolute) = '20% fewer' (relative). Small samples are noisy (coin flips: 17% vs 4 in 100,000). Under an hour of training improved forecasts 6-11% (Chang 2016); little evidence on life outcomes. Do this week: rewrite one claim as 'out of 1,000 people, X vs Y'.