Skip to content
Omnikalc

Rasch model calculator

Enter the ability of a person and the difficulty of three items, both in logits, to get the probability of a correct answer on each item, the expected score, the information of the test and the standard error of the ability.

Updated on
Your inputs
logits

Value out of range

logits

Value out of range

logits

Value out of range

logits

Value out of range

Results

Probability of success on item 1

0.8176

Probability of success on item 2
0.6225
Probability of success on item 3
0.2689
Expected score
1.709 / 3
Information of the test
0.5808
Standard error of the ability
1.312 logits
Imprecise: add items close to the ability of the person
Best matched item (smallest gap)
2

The Rasch model calculator gives the probability that a person answers each item correctly, the expected score, the information of the test and the standard error of the ability, on the logit scale.

How the calculation works

  • Enter the ability of the person in logits.
  • Enter the difficulty of three items in logits.
  • The calculator gives the three probabilities, the expected score, the information, the standard error and the best matched item.

Example

A person has an ability of 0.5 logits and meets items with difficulties of −1, 0 and 1.5. The probabilities are 0.8176, 0.6225 and 0.2689, so the expected score is 1.709 out of 3. The items give information of 0.149, 0.235 and 0.197, a total of 0.5808, and the standard error is 1 / √0.5808 = 1.312 logits. The middle item is the best matched.

Why it matters

The calculator helps students in psychometrics and education, test designers who calibrate items, and teachers who want to understand how item difficulty relates to results.

Good to know

Real calibrations need response data from many people and a software that estimates ability and difficulty at the same time. This calculator applies the model once you have estimates. The Rasch model assumes equal discrimination for all items, so that the item curves do not cross; the two-parameter model relaxes this. A short test of three items is only an illustration, and real tests have 20 or more items to get a standard error under 0.5 logits. The scale origin is arbitrary and usually set by centering the items on zero. Always report the scale choice along with the results.

The item difficulty and discrimination calculator uses classical test theory, the Cronbach alpha calculator measures reliability, the logistic regression calculator gives probabilities from coefficients and the sigmoid function calculator evaluates the same curve.

Frequently asked questions

What is the Rasch model?

It is the simplest item response model. The chance that a person with ability θ answers an item of difficulty b correctly is P = 1 / (1 + e^−(θ − b)). Ability and difficulty sit on the same scale, in logits, and only their difference matters.

What does a logit mean?

A logit is the natural log of the odds. When ability equals difficulty, the chance is 50 %. One logit above gives 73 %, two logits 88 %, and one logit below 27 %. An item that is a logit harder multiplies the odds of success by 0.37.

Why is the information important?

Each item gives information P × (1 − P), which peaks at 0.25 when the item matches the person. The standard error of the ability is one over the square root of the total information, so well-targeted items measure better than very easy or very hard ones.

Was this calculator helpful?

Thanks for your feedback!

Results are estimates provided for information purposes. They do not replace professional advice.