Objective: Work through a set of intelligence problems and make the judgment the material actually supports. Every problem is generated fresh, so the test is different each time you take it.
The six sections:
- Source Evaluation — reports are graded for source reliability (A to F) and information credibility (1 to 6). You decide which report deserves most weight, spot reports that only appear to corroborate each other because they share one original source, and notice when a strong rating on the source is being borrowed by weak second-hand content.
- Evidence vs Inference — you take a statement about a report and classify it as something stated in the report, a conclusion that follows from what is stated, or an assumption the report does not support. Most analytic errors start here.
- Competing Hypotheses — two explanations are laid out with a set of evidence items. You identify which item actually discriminates between them, which explanation the evidence favors, and what new finding would weaken a hypothesis. When every item fits both explanations, the correct answer is that the evidence does not separate them.
- Timeline Reconstruction — a travel-time table and a movement log are given. You find the entry that cannot be true, or work out which stop was possible inside a confirmed window.
- Link Analysis — a contact list describes a network. You find the cut-out linking two groups, the single link whose loss would split the network, or the one person in contact with a given set of individuals.
- Analytical Judgment — three things an analyst has to do once the evidence is in. Pick the estimative judgment the reporting supports, from almost certainly through to insufficient basis, where overstating and understating are both errors. Choose the collection step that would actually resolve the key uncertainty rather than adding background. And pick the written assessment that separates what is known from what is inferred, keeps a viable alternative on the table, and states confidence at the level the evidence can carry.
Controls: Click an answer, or press the matching number key. The test advances automatically once you answer. If you set a per-question time limit, an unanswered question is marked incorrect when the timer runs out.
Settings:
- Test Length — 12, 18, or 24 questions, split evenly between the six sections.
- Difficulty — Challenging affects all six sections: more reports with closer gradings, subtler statements to classify, more evidence items per hypothesis set, larger networks and longer movement logs, near-miss collection options, and subtler confidence-calibration profiles across the full confidence scale.
- Focus Area — under Advanced Settings, restrict the whole test to a single section when you want to drill one skill.
Review: Missed questions are listed at the end of the test with the answer you gave, the correct answer, and the reasoning behind it. Nothing is revealed during the scored run.
Scoring: Each question is worth one point. The skill breakdown is the useful part of the result if you are practicing, because the six sections fail in different ways: source evaluation errors come from treating repetition as corroboration, evidence errors come from filling gaps with assumption, and judgment errors come from stating more confidence than the reporting can carry, collecting more of what you already have, or writing an assessment that quietly buries the alternative.