Spot the Statistical Lie

Find one statistic in your feed and interrogate it with four questions.

First: what is the distribution? "Average household income rose 5%" could hide a 30% jump at the top and no change at the bottom.

Second: what is the sample? "90% of our customers are happy" means little if only happy customers answered. Ask for the response rate.

Third: what is missing? "Sales up 200%" with a truncated y-axis might be a $2 rise on a $1 base.

Fourth: is there a Simpson's-paradox risk? Unequal subgroups can hide contradictions in the aggregate.