Test Item Analysis Calculator
Analyse exam questions from a mark sheet to find the difficulty index, discrimination index and point-biserial correlation, identifying questions that are not measuring anything.
Last reviewed by the Radiatus Cloud team
Need this done properly for your business?
Radiatus delivers secure cloud, DevOps & compliance engineering.
A question everyone answers correctly measures nothing
An item answered correctly by every candidate contributes no information about who knows the material, because it cannot separate anyone from anyone else. The same is true of one everybody gets wrong. Difficulty around fifty to sixty percent produces the most discrimination, which is why a well-constructed test contains a spread rather than only easy questions. Very easy items still have a place at the start of a paper, for confidence, and they do not contribute to the measurement.
Discrimination matters more than difficulty
The discrimination index compares performance on an item between the strongest and weakest candidates. A positive value means the students who did well overall also did well on that item, which is what a question measuring the same thing as the rest of the test looks like. A negative value means the weaker students did better on it, which nearly always indicates a flawed question: an ambiguous stem, a miskeyed answer, or a distractor that is arguably correct.
A negative discrimination is a defect, not a hard question
It is worth separating the two. A hard item has a low difficulty index and positive discrimination: few get it right, and those who do are the stronger candidates. A defective item has negative discrimination, and no amount of difficulty explains why the weakest students outperform the strongest on it. Those items should be reviewed and usually removed from scoring before results are issued.
Related tools
- Security Term Translator — Translate complex security jargon into business language.
- Security Myth vs Reality — Fact-check common security myths.
- Breach News Impact — Contextualize major breaches into understandable analogies.
- Config Safety Analyzer — Paste a config snippet to check for obvious safety issues.
Frequently Asked Questions
What is the difficulty index?
The proportion of candidates answering correctly, so a higher number means an easier item. The name is confusing for that reason: 0.9 is easy and 0.2 is hard.
What discrimination value is acceptable?
Above 0.4 is good, 0.3 to 0.4 is reasonable, 0.2 to 0.3 is marginal and below 0.2 is weak. Anything negative indicates a defective item rather than a difficult one.
What causes negative discrimination?
Nearly always a flaw: an ambiguous stem, a miskeyed answer, a distractor that is arguably correct, or a question testing something the rest of the test does not. Strong candidates over-thinking a poorly worded item is a common mechanism.
What is the point-biserial correlation?
The correlation between performance on one item and total score. It uses every candidate rather than just the top and bottom groups, so it is more stable on small cohorts than the discrimination index.
Should I remove bad items from scoring?
Items with negative discrimination, usually yes, after checking the key. Removing items after the fact changes everyone's mark, so the decision should be documented and applied consistently rather than case by case.
Privacy & Security
Everything runs in your browser; nothing is uploaded.
How to Use
Paste a grid of student answers marked right or wrong to analyse each question.
Disclaimer: This tool is provided "as is" without warranty of any kind. Results are for educational and utility purposes.