Evaluating Cumulative Standard Errors of Chained Linear Equating for a High-Stakes Clinical Performance Examination

• In Event: 72.025 - Psychometric Issues and Performance Over Time

Abstract

Chained Levine observed score equating under a nonequivalent-groups with anchor test (NEAT) design was used to link scores from new test forms to the base form for a high-stakes objective structured clinical examination (OSCE) for score comparison overtime. The concern was that after a series of links, equating error may accumulate to a degree that affects score comparability. A bootstrap method was used with real examinee data to evaluate cumulative standard errors (SEs) of chained equating. Results indicated that: (1) SEs accumulated slightly as the number of links increased; (2) SE accumulation was smaller near the mean than towards extreme scores; (3) the accumulated SEs were within the standard error of measurement in the typical observed score range.

Authors

• Fang Tian, Medical Council of Canada

• Andrea Gotzmann, Medical Council of Canada

• Sirius Qin, Medical Council of Canada

• Liane N. Patsula, Medical Council of Canada

• André F. De Champlain, Medical Council of Canada