Logic
1,661 questions we hold — every difficulty
421 hard questions models are actually tested on · 36 extra-hard ceiling
The two counts describe different parts of the question bank. The pool includes easy, difficult, and hard questions so coverage can be inspected, but those lower-difficulty questions are not used in the sampled test. Each attempt draws from the hard set: twenty questions per subject when the cell allows it, plus ceiling. 1,661 distinct skills. Live prompts and answers stay private. The examples below show the structure of hard questions without exposing the live test. Words we use
Style samples cycle at random. Each card illustrates the structure of a hard question: a setup followed by the task. These examples are not used in the live test. Where the displayed information is sufficient to determine the result, the answer and a short explanation are shown as well.
Data analysis
Research
Perception
Coding
01
Subject × difficulty
Four fields meet the hundred-question calibration cells. Coding meets twelve per calibration cell and has a hard set large enough for a twenty-question draw. Easy, difficult, and hard stay in the pool and are not asked. Skill families group unique constructs for coverage.
02
Subject × skill cluster
Clusters are public tags for what the question tests. Cells are counts. Empty is zero. Titles, prompts, and answers stay off this page.