Here are the core steps needed for practical adaptive testing with the Rasch model.
0. Request next candidate: Set D=0, L=0, H=0, and R=0.
1. Find next item near difficulty (D).
2. Set D at the actual calibration of that item.
3. Administer that item.
4. Obtain a response.
5. Score that response.
6. Count the items taken: L = L + 1
7. Add the difficulties used: H = H + D
If response not correct,
8. Target next item difficulty: D = D - 2/L
If response correct,
9. Count right answers: R = R + 1
10. Target next item difficulty: D = D + 2/L
If not ready to decide to pass/fail,
11. Go to step 1.
If ready to decide pass/fail,
12. Calculate wrong answers: W = L - R
13. Estimate measure: B = H/L + loge(R/W)
If W = 0 then B = H/L + loge(R-0.5 / W+0.5)
If R = 0 then B = H/L + loge(R+0.5 / W-0.5)
For a more precise estimate of B, see
Estimating Measures with Known Item Difficulties
14. Estimate Error: S = sqrt[L/(R*W)]
If W = 0 then S = sqrt[L/(R-0.5 * W+0.5)]
If R = 0 then S = sqrt[L/(R+0.5 * W-0.5)]
15. Compare measure B with pass/fail standard T.
16. If (T - S) < B < (T + S), go to step 1.
17. If (B - S) > T, then pass.
18. If (B + S) < T, then fail.
20. Go to step 0.
Practical adaptive testing. Wright BD. … Rasch Measurement Transactions 2:2 p.24
Practical adaptive testing. Wright BD. … Rasch Measurement Transactions, 1988, 2:2 p.24
The URL of this page is www.rasch.org/rmt/rmt22g.htm
Website: www.rasch.org/rmt/contents.htm