IRT measurement models

Question 1

One-parameter logistic model (1PL / Rasch Model)

Answer 1

Response to dichotomous item (yes/no right/wrong etc). Probability of answering correctly, based on knowledge of theta. Depends on item difficulty (left= easy item right=hard)

Question 2

s-shaped functions

Answer 2

The probability on the Y axis is notes with an S-curve to show the probability of answering an item correctly for each participant.

Curve can change based on the difficulty of the item.

the middle of the curve represents the discrimination.

Question 3

2 parameter logistics model (2PL)

Answer 3

Includes two parameters;
Trait level (theta) depends on
- item difficulty
- item discrimination (steepness of the curve, the steeper the better the discrim. )

Extension of the Rasch model but includes the discrimination.

Question 4

3 parameter logistic model (3PL)

Answer 4

Same as 2PL but adds correction for the guessing score. Adds c to correct the probability you guess correctly.

Question 5

Graded response model

Answer 5

Polytomous items (Likert). Multiple response options instead of binary. Each item has its own curve. The probability decides the position of the curve.