How IQ tests are standardized

Standardization means administering a test the same way every time and converting raw scores using norms from a large, representative sample. Test developers recruit a standardization sample matched to the population on characteristics such as age, sex, education and region, administer the test under controlled conditions, then build score tables so any future test taker can be compared with people of the same age.

If one person gets extra time, clearer instructions or a quieter room, their score isn't comparable to the norms. That's why professional test manuals specify wording, timing, materials and how to handle questions. Departures from the procedure are recorded, because they change what the score means.

Average measured performance rose substantially through the twentieth century, a pattern documented in a meta-analysis covering more than a century of data. Because scores are relative to norms, unchanged norms would gradually inflate everyone's scores. Renorming resets the average to 100 for the current population, which is why a person can score slightly lower on a newly normed test than on an older one.

Most online tests have no documented standardization sample and no control over conditions. Some compare you with everyone who has taken the test on that site, a self-selected group, which is not a population norm. That's the core reason an online number can't be treated as an IQ in the technical sense.