Five Common Tricks

Watch for these in every data-driven argument:

• Cherry-picking — using only the data that supports your conclusion
• Misleading averages — mean, median, and mode can tell completely different stories
• Relative vs absolute numbers — "doubles your risk" sounds scary until you learn the risk went from 1 in a million to 2 in a million
• Correlation vs causation — ice cream sales and drownings both rise in summer, but ice cream does not cause drowning
• Survivorship bias — studying only the successes and ignoring all the failures