The Research on Anchor Ratio of Test Equating Design

摘要

Test equating is necessary for large scale item bank constructing. Before the actual test equating, not only time consuming, the consumption of wealth, but also the safety of the item may be endangered, particularly greater risk for anchor of repeated use, so the anchor ratio (R) of test equating design is expected smaller. Currently reported result in the country is: R is between 1 / 4 and 1 / 3 of the test length, test equating to be valid. So as not to affect equating accuracy ,this paper discusses R of different lengths . Monte Carlo (MC) simulation results show that, R can change with changes in test length, such as about test length for the 600,300,200,100 items, R can be reduced to 1 /15,1 /12, 1 / 10, 1 / 5; when the test length is 40 items, R at least must be to 1 / 3.