Frost sets out to teach statistical reasoning to readers who mostly encounter numbers and charts produced by someone else and want to judge whether they can be trusted, and he does it with explanation rather than formal derivation.

An opening section pits statistical evidence against anecdote, using a weight-loss supplement to show why testimonials mislead: the people whose results were unremarkable never come forward, and a self-selected sample tells you nothing about anyone else. Observational studies, which measure many relevant variables and model the contribution of each, are distinguished from randomized controlled trials, which alone support causal conclusions.

The book then splits in two. The first half covers working tools — kinds of data, quantitative and categorical, continuous and discrete; histograms, boxplots and contingency tables, with cautions about how each can mislead; central tendency, variability, percentiles and correlation; and probability distributions, dwelling on the normal curve and standard scores. The second turns to inference from sample to population.