Test Item Analysis Calculator
Paste your class's answers to find which test questions worked: difficulty and discrimination for each item, which distractors fooled the right students, and the whole test's reliability.
Difficulty p is the share of students who got a question right (0.75 = 75%). Discrimination D = p in the top 27% minus p in the bottom 27%: if 9 of 10 strong students and 3 of 10 weak students get it right, D = 0.9 − 0.3 = 0.6, a question that separates them well. A negative D usually means a wrong key or a misleading question.
Your data
Upper and lower groups
Copy the rows from a spreadsheet, or type them: "Maya A B C D" or "ABCD". A name in the first column is optional; leave a blank or "-" for no answer.
One letter per question, e.g. "A B C D" or "ABCD".
Whole test
- Students: 12 · Questions: 6
- 3 students in each group
- Mean score 3.00 of 6 (standard deviation 1.96)
- Reliability (KR-20): 0.77 · Acceptable for a classroom test.
| Q | Difficulty p | Upper | Lower | Discrimination D | Level | Discrimination | Suggestion |
|---|---|---|---|---|---|---|---|
| 1 | 0.67 | 1.00 | 0.33 | 0.67 | Moderate | Very good | Keep |
| 2 | 0.50 | 1.00 | 0.00 | 1.00 | Moderate | Very good | Keep |
| 3 | 0.42 | 1.00 | 0.00 | 1.00 | Moderate | Very good | Keep |
| 4 | 0.25 | 0.67 | 0.00 | 0.67 | Hard | Very good | Keep |
| 5 | 0.58 | 1.00 | 0.33 | 0.67 | Moderate | Very good | Keep |
| 6 | 0.58 | 1.00 | 0.00 | 1.00 | Moderate | Very good | Keep |
Answer choices by group
Q 1
| Choice | All | Upper | Lower | Distractor |
|---|---|---|---|---|
| A | 8 | 3 | 1 | key |
| B | 2 | 0 | 1 | works |
| C | 1 | 0 | 0 | doesn't work |
| D | 1 | 0 | 1 | works |
Q 2
| Choice | All | Upper | Lower | Distractor |
|---|---|---|---|---|
| A | 2 | 0 | 2 | works |
| B | 6 | 3 | 0 | key |
| C | 2 | 0 | 0 | doesn't work |
| D | 2 | 0 | 1 | works |
Q 3
| Choice | All | Upper | Lower | Distractor |
|---|---|---|---|---|
| A | 3 | 0 | 1 | works |
| B | 3 | 0 | 1 | works |
| C | 5 | 3 | 0 | key |
| D | 1 | 0 | 1 | works |
Q 4
| Choice | All | Upper | Lower | Distractor |
|---|---|---|---|---|
| A | 3 | 1 | 1 | doesn't work |
| B | 3 | 0 | 1 | works |
| C | 3 | 0 | 1 | works |
| D | 3 | 2 | 0 | key |
Q 5
| Choice | All | Upper | Lower | Distractor |
|---|---|---|---|---|
| A | 7 | 3 | 1 | key |
| B | 3 | 0 | 1 | works |
| C | 1 | 0 | 0 | doesn't work |
| D | 1 | 0 | 1 | works |
Q 6
| Choice | All | Upper | Lower | Distractor |
|---|---|---|---|---|
| A | 1 | 0 | 0 | doesn't work |
| B | 2 | 0 | 2 | works |
| C | 7 | 3 | 0 | key |
| D | 2 | 0 | 1 | works |
Bands are common rules of thumb, not standards: difficulty below 0.30 is hard and above 0.70 easy; discrimination of 0.40+ is very good, 0.30–0.39 good, 0.20–0.29 marginal, below 0.20 poor. A distractor works when at least 5% choose it and more lower-group than upper-group students do.
How to use the item analysis calculator
- 1Choose answer letters with a key (multiple choice), or scores (1/0, or points for written questions).
- 2Paste the rows from your spreadsheet: one student per row, one question per column. Names are optional.
- 3Type the answer key, or the points each question is worth.
- 4Choose the top and bottom 27% (the usual rule), or halves for a small class.
- 5Read the table: questions to keep, revise or replace, which distractors work, and the test's reliability. Download the CSV or print the report.
When this is useful
After a unit test
Find the questions everyone missed and the ones that didn't separate students who understood from those who didn't.
Fixing a question bank
Keep the strong questions, rewrite the marginal ones and replace the rest before reusing a test.
Checking the key
A negative discrimination flags a question where the strongest students chose a different answer, often a key error.
Department reviews
Share the CSV so colleagues can review a common exam with the same numbers.
Teacher training assignments
Show the full calculation for a coursework item analysis, with the groups and formulas stated.
Examples to try
A question most students got right
p = 0.92: easy. Fine as a warm-up, but it tells you little about who understands.
A question that separates well
Upper group 0.90, lower group 0.30: D = 0.60, very good. Keep it.
A distractor nobody chose
Option D picked by 0 students: it doesn't work. Replace it with a common wrong answer.
A negative discrimination
Upper 0.20, lower 0.60: D = −0.40. Check the key, then the wording.
Written questions
Scores of 0–4 on an essay question: p = average ÷ 4, and reliability uses Cronbach's alpha.
Tips for better results
Use enough students
With fewer than about 20 students, one answer moves the numbers a lot. Use halves instead of 27% groups and read the results as hints.
Easy isn't bad
A very easy question can be exactly right for checking a basic skill. Look at discrimination before discarding it.
Keep the same rows
Copy whole rows from the spreadsheet so every student's answers stay in the same question columns.
Rules of thumb only
The bands for difficulty, discrimination and reliability are common conventions. Your school or course may use other cut-offs.
How the item analysis calculator works
Each student's total is the number of correct answers (or points). Students are ranked by total; the top and bottom 27% (or halves) form the upper and lower groups. For each question, difficulty p is the proportion correct in the whole class (or the mean score divided by the maximum), and discrimination D is p in the upper group minus p in the lower group.
For multiple choice, each option is counted in each group. A distractor is marked as working when at least 5% of students choose it and more students in the lower group than in the upper group do.
Reliability uses KR-20 when every question is scored right or wrong, and Cronbach's alpha when questions carry points. Both use the spread of total scores; values from about 0.70 are usually considered acceptable for a classroom test.
Private by design
- No account and no sign-up. Just open it and start.
- Everything runs on your device, which keeps the answers, scores and names you paste private.
- Nothing you create is uploaded. No values, names, scores, or generated content are sent to our servers. If you're signed in, My Classeem's Recent Activity notes which tools you used — never what you did in them.
- We use Google Analytics for anonymous, aggregate measurement: page views and a few usage events, such as a worksheet being printed or a game round starting or finishing (with the mode played). It never receives what you enter — no answers, scores, names, or generated content.
Frequently asked questions
What is the difficulty index?
The proportion of students who answered correctly, from 0 to 1. 0.75 means 75% got it right. Higher numbers mean easier questions; around 0.30 to 0.70 is often called moderate.
What is the discrimination index?
How much better the strongest students did than the weakest on one question: p in the upper group minus p in the lower group. From 0.40 is usually called very good, 0.30 to 0.39 good, 0.20 to 0.29 marginal and below 0.20 poor.
Why 27%?
Comparing the top and bottom 27% is a long-standing convention (Kelley) that balances group size with how different the groups are. For small classes, many teachers compare the top and bottom halves instead.
What does a negative discrimination mean?
More weak students than strong students got it right. Check the answer key first; if it is correct, the question may be ambiguous or rewarding guessing.
Can I analyze essay or open questions?
Yes. Choose scores and enter the points each student earned, plus the points each question is worth. Difficulty becomes the average score divided by the maximum.
Are my students' answers uploaded?
No. Everything is calculated in your browser, and the CSV is created on your device.
More classroom tools
Test Grade Calculator
Turn the number wrong into a percentage and letter grade, with a printable chart for the whole test.
Grade Calculator
Enter scores to see a current grade as a percentage and a letter, with weighted categories or a simple average.
Quiz Maker
Make a printable classroom quiz: multiple choice, true/false, and short answer, with an answer key.
Always free · No account needed · Nothing you type leaves your device unless you choose to save it.