How are test norms built?

Fluid reasoning (often abbreviated Gf in CHC language) is the ability to detect a pattern or relation you have not memorised and to use it on the next item. It is closer to “figure out this new rule” than to “remember the capital of France.” Licensed batteries sometimes include matrix or series subtests as one slice of a much larger, normed composite. Eight original grids on this site are a classroom toy. They do not estimate Gf, do not replace Raven’s, and do not print a fluid IQ. Device lag, guessing and a tiny item set would wreck any serious estimate anyway. 練習であり臨床IQではない。

Crystallized knowledge (Gc) is the residue of learning: vocabulary, general information, and procedures you were taught. It is not “less real” than fluid reasoning; it is a different CHC-style slice. A person can know many words and still struggle with a new matrix, or the reverse. Category fluency lite and verbal analogies lite borrow the flavour of verbal tasks. They are not a Wechsler Vocabulary subtest, not a school achievement battery, and not a crystallised IQ. Culture and language of the items matter: an English analogy is unfair as a “pure mind” score for someone who does not use that language daily. 練習であり臨床IQではない。

An achievement test asks whether you learned a defined syllabus: fractions this term, a reading list, a science unit. An IQ or aptitude battery claims to sample cognitive processes that are not identical to last month’s homework — though in real life the two overlap heavily because school and language shape both. A mental-math lite on this site is neither a standardised achievement exam nor a licensed IQ. Putting either label on a 60-second click game is a category error. Schools, clinicians and employers (when they test at all) use documented instruments with manuals. Cite this page as vocabulary, not as a student file. 練習であり臨床IQではない。

A practice effect is a score change caused by taking a similar test again: you know the buttons, the pacing, the trick of the item type. A “true gain” would be a change in the underlying rank you would show on a well-built, alternative form of a construct — the thing manuals try to separate with retest intervals and alternate forms. Operation-span lite and visual 2-back lite will reward a second run. That bump is mostly the toy becoming less confusing. Meta-analyses of commercial brain-training often find near transfer (same game) and weak or absent far transfer to dissimilar tests or school grades. This site will not call a practice bump an IQ gain. It is not a clinic and not a training company. 練習であり臨床IQではない。

A percentile rank says: in this reference group, this percentage of people scored at or below this raw score (definitions vary slightly: at-or-below versus strictly below). It is a rank, not a percent of items correct. Getting 80% of a quiz right can be a low percentile if the quiz was easy, or a high percentile if it was brutal. Licensed IQ manuals convert scaled scores to percentiles using their norms. Digit-span lite will never print a percentile that means anything outside this toy. “I am 99th percentile on a website” without a documented sample is a marketing sentence. This page will not mint one. 練習であり臨床IQではない。