Sec.3-Ch.1-Subsec.5：Understanding Survey Weights in NHANES

How complex sampling affects estimates, uncertainty, and real-world interpretation

Some people simply take the sample they have and start calculating indicators right away. That’s understandable—but if you ignore the sampling design and the survey weights, your results can end up pretty far from the truth.

NHANES is a classic example. It doesn’t sample people purely at random. Some subgroups are intentionally oversampled so that we actually have enough data to analyze them. The problem is, if you analyze the raw, unweighted data, those groups end up having way more influence than they should.