Computerized Adaptive Testing — Item Response Theory Simulator

Modern computerized tests don't ask every candidate the same fixed list of questions — they pick each next question in real time, choosing whichever item will reveal the most about the test-taker given what's been learned so far. This 2D simulator plots a 50-item bank across a difficulty/information map and runs a genuine computerized adaptive test against it: every item follows the two-parameter logistic IRT model, the engine re-estimates the examinee's ability with a Newton-Raphson maximum-likelihood update after each response, and the maximum-information selection rule steers the next question toward the difficulty level that will shrink the standard error fastest. Set the hidden true ability, step through items one at a time or auto-run the test, and use "Run Comparison" to race the adaptive engine against a fixed-order test on the same bank — the convergence chart shows, item by item, how much faster the adaptive θ̂ estimate locks onto the true ability.