【摘要】 在无锚情况下，国外学者提出可利用被试背景信息变量对非等组的被试进行最小鉴别信息加权构建伪等组实现分数连接。为探讨和分析伪等组等值在我国测验情境中应用的准确性，采用实证数据，对比分析了伪等组等值与传统锚题等值的结果。结果显示，在被试背景信息变量与待等值测验分数的相关达到0.75左右时，伪等组等值所产生的等值误差接近传统锚题等值，其等值分数与传统锚题等值分数的对比结果也显示，伪等组等值不存在等值重要差异，即二者的差异不会带来实际影响。结果表明，在一定测验情境下，伪等组等值可以成为传统等值的有效替代。未来研究可对伪等组等值在不同测验情境中应用的准确性进行更多验证。更多还原

【Abstract】 In the case of a nonequivalent group design with no common items, pseudo-equivalent groups can be formed by weighting samples of examinees through adjustment by minimum discriminant information with background information variables to provide an approach to linking test forms. In order to explore and analyze the accuracy and feasibility of the pseudo-equivalent groups equating in China’s real testing programs, this study applied this new method to a testing program in China. Results were analyzed and compared to those from the non-equivalent anchor test equating. It showed that when the background variable had a correlation of about 0.75 with test scores, equating with pseudo-equivalent groups can produce a similar equating error to traditional anchor test equating, and no difference matters between them, which means differences between the equating results from pseudo-equivalent groups and anchor test would not bring real impact. Thus, pseudoequivalent groups can be an alternative to traditional equating methods under certain testing situations.However, its accuracy should be further verified in more testing situations in future studies.更多还原