March 2026 · Assessment
Item analysis is the process of evaluating individual test questions after administration. It tells you which items worked well, which were too easy or too hard, and which distractors were ineffective. Regular item analysis dramatically improves your tests over time.
Facility value (FV) = number of students who got the item right ÷ total students. FV of 0.90 means 90% got it right (very easy). FV of 0.20 means only 20% got it right (very hard). Target FV: 0.30-0.70 for most items. Items below 0.20 may be flawed or testing content not yet learned. Items above 0.90 don't discriminate and could be removed.
Calculate FV for every item on your exam. Create a difficulty profile: if most items are above 0.80, your exam is too easy. If most are below 0.40, it's too hard. Aim for a bell curve centered around 0.50-0.60.
The discrimination index (DI) measures how well an item separates strong students from weak students. Divide students into top 27% and bottom 27% by total test score. DI = (top group correct - bottom group correct) ÷ group size. DI above 0.40 = excellent discrimination. DI 0.20-0.39 = acceptable. DI below 0.20 = poor (revise or remove). Negative DI = the item is flawed (weak students got it right more often than strong students).
Items with negative DI need immediate investigation: Is the answer key wrong? Is the question ambiguous? Does it test something irrelevant? Negative DI items should be removed from scoring and rewritten for future tests.
For multiple choice items, analyze each distractor (wrong option). A good distractor attracts at least 5-10% of students. Distractors chosen by nobody are "dead" — they're obviously wrong and need replacement. Distractors chosen more than the correct answer indicate a problem: the correct answer may be wrong, the question ambiguous, or the distractor confusingly close to correct.
After analysis, categorize items: Keep (good FV and DI), Revise (acceptable FV but poor DI, or good DI but extreme FV), Remove (negative DI, extreme FV, and dead distractors). Build an item bank of tested, validated questions. Over 3-4 test cycles, your exam quality will improve dramatically.
No. A spreadsheet (Excel/Google Sheets) is sufficient. Enter 1 for correct, 0 for incorrect per student per item. Calculate FV with AVERAGE and DI by comparing top/bottom groups. There are also free online tools like TAP (Test Analysis Program) that automate the process.
Ideally 30+ students for meaningful statistics. With fewer students, FV is still useful but DI becomes unreliable. For small classes (10-15), focus on FV and qualitative analysis of student responses rather than statistical indices.