Equating

Abstract

When different tests are administered, the results from the tests are not directly comparable. A process called Equating is needed for comparing results from different tests.

Appendices

Discussion Points

• (1) Discuss the relative merits of the shift, anchoring and joint calibration methods of equating. What criteria would you devise in order to choose among these three equating methods?

• (2) Discuss how the presence of equating error would affect the results from a linking study between two tests with a set of common items. What are some practical steps during the instrument design or test preparation stage that can minimize potential equating error later on?

Exercises

Q1. The following table shows estimated item parameters for 20 link items from two tests separately calibrated. If Test 2 needs to be placed on Test 1 scale, compute the equating transformation using the shift method. Also compute the equating error

Link item	Test 1	Test 2
1	0.66	0.98
2	−0.64	−0.21
3	0.80	1.20
4	−1.44	−1.08
5	−0.59	−0.44
6	−0.86	−0.46
7	1.41	1.91
8	−0.74	−0.59
9	−1.41	−1.26
10	−0.15	0.22
11	−1.11	−0.84
12	−0.69	−0.34
13	−0.85	−0.55
14	−0.40	−0.15
15	0.64	1.10
16	0.24	0.49
17	0.17	0.46
18	−1.07	−0.85
19	−0.15	0.31
20	−0.73	−0.67