Every item on this test is a figure or a number sequence. There is no vocabulary section, no general-knowledge section, no word problems. That is a deliberate trade.
What a vocabulary question actually measures
Asking for the antonym of generous measures two things at once: how well someone reasons, and how much English they have been exposed to. For a native speaker who reads widely, those are hard to separate. For someone who learned English at school in another country, the second swamps the first.
The same applies to general knowledge and to arithmetic drill. A question about percentages measures whether you were taught percentages recently.
The three families used instead
Matrix reasoning. A three-by-three grid where every row and column obeys the same rules. Work out the rules, pick the missing cell.
Number series. A sequence built by one hidden operation, sometimes two alternating. Nothing beyond adding and multiplying small numbers — the difficulty is in spotting the rule, not in the arithmetic.
Figural analogy. The first figure turns into the second by some change. Apply the same change to the third.
All three are rules waiting to be spotted. None of them can be studied for in the way a vocabulary list can.
What this costs
Non-verbal reasoning is not all of intelligence, and a test that drops verbal ability is measuring something narrower than a full assessment would. A clinical battery takes hours and includes verbal subtests for good reason.
This is a 20-minute screening test, not a clinical assessment. Within that limit, measuring one thing cleanly beats measuring several things badly.
Items are generated, not copied
The items in Raven's Progressive Matrices, WAIS and Cattell's CFIT are all under copyright and cannot be reproduced. Every question here is generated from rules instead — which also means difficulty is controlled by how many attributes vary at once, rather than assigned by hand on a hunch.