Paper Summary
Share...

Direct link:

The Effect of Anchor Construction on Test Score Equating

Mon, April 16, 10:35am to 12:05pm, Westin New York at Times Square, Floor: Fourth Floor, Gershwin II Room

Abstract

Sinharay and Holland suggested that the requirement for an anchor block which was built to represent a miniature form of the total test in terms of content and statistical specifications may be too restrictive. A simulation study was conducted to investigate the effect of content and statistical representativeness of an anchor on test score equating in K-12 related examinations. The impact was evaluated and compared with respect to SEE, Bias and RMSE. The results shows that midtest perform as well as or better than traditional minitests for most cases and appear to be relatively robust to large differences in group ability. The content representativeness tends to have more impact on equating accuracy than the statistical representativeness of anchor test.

Authors