This article tackles the problem of what should be done with real textual data that are contaminated by errors of recording, particularly when the data contain words that are misspelt, unintentionally or otherwise. (Contains 5 tables and 2 figures.)

Suggests using simulated data to teach students about the nature and use of statistical tests and measures. Observes that simulated data contains built-in pure relationships with no poor response rates or coding or sampling errors. Recommends suitable software. Includes information on using data sets, demonstrating statistical principles, and…