real data is never perfectly normal, but the model is useful. This "all models are wrong, but some are useful" insight prepares students for the CLT (Ch. 11) and assumption checking throughout the inference chapters.