Scored Items Versus Items You Answer

Every section mixes two kinds of questions together: the ones that count toward the score, and a smaller batch that doesn't. The uncounted ones are field-test items. The people who build the test need to know how a new question performs before trusting it on a real form, so they slip it into a live administration and watch how test-takers handle it, without using the results against anyone. That's how 171 questions on a form become a score built from only 142 of them. The remaining 29 are split across the sections: 10 in English, 4 in math, 9 in reading, 6 in science.

The mechanism depends on those 29 being invisible. If a field-test question looked different, sat in a predictable spot, or felt out of place, test-takers would learn to coast on it or skip it, and the data it produces would be worthless. So these items are written to match the style, difficulty, and formatting of scored ones, and they're scattered through each section rather than clustered at the end. There's no tell. A question that feels oddly hard or oddly easy is just as likely to be scored as not.

This sits right next to how the raw score gets built: only correct answers on the 142 scored questions count, so if someone later learns that a few of their correct answers landed on field-test items, the scored tally doesn't change at all. It also explains why timing guidance always treats every question as if it matters. Since nobody can tell which ones are live and which are trial runs, the only reliable strategy is full effort straight through.