Item Discrimination Simulator

Upper/Lower Method in Classical Test Theory (CTT)
1

Form the groups using the corrected total score X(−i)

First, respondents are ordered by the corrected total score X(−i), that is, the test score excluding the item being analysed. The 27% with the lowest X(−i) form the Lower group, and the 27% with the highest X(−i) form the Upper group.
Lower
27%
Middle group
46%
Upper
27%
Lower X(−i) Higher X(−i)
Important: X(−i) is used to define the groups. Discrimination is calculated for that specific item.
2

Examine responses to a specific item

Item being analysed: Item i
The two proportions below refer to the same item.

Upper group

27% with higher X(−i)
pU = 0.80
Proportion correct in Upper 80%
0.000.501.00
Correct responses: 80 Incorrect responses: 20

Lower group

27% with lower X(−i)
pL = 0.30
Proportion correct in Lower 30%
0.000.501.00
Correct responses: 30 Incorrect responses: 70
Correct responses in Upper
Correct responses in Lower
Incorrect responses

Separation between groups for Item i

Each marker's position represents its proportion correct. As the markers move farther apart, the magnitude of the difference between the groups increases.
Upper = .80
Lower = .30
0.000.250.500.751.00
D = +0.50
3

Calculate item discrimination

Di = pU − pL The proportions correct for the two groups are compared for the same item.
Di = 0.80 − 0.30 = 0.50
D for Item i = +0.50
The Upper group answered this item correctly more often than the Lower group.