摘要
研究对某次大型人事考试的主观评分结果进行分析。结果表明,各题上考生得分的离散程度偏小,评分者所评分数的分布比较集中且处于低分段,体现出评分者打分时相当保守,同时分数全距缩小也降低了主观性试题的效用。各题均值和标准差都随评分时间的增长而有所减小,体现了评分保守是一个"习得"的过程。为控制这种现象的发生,在不降低评分一致性的前提下,进一步提升评分的准确性,可通过优化误差监控的指标,插入标准卷,以及采用分步增值评分模式进行流程控制。
This study analyzed one large - scale personnel selection writing test. The results showed that the SD of scores on each item was small,which meant that the examinees＇ abilities could not be well - distinguished. The raters tended to give lower and concentrate scores due to their conservative strategy. Meanwhile, the shortened range of scores lowered the efficacy and effectiveness of those subjec- tive items. It is also found that the means and Sl）s decreased along with the rating progress ,which might imply that the conservative sco- ring was an acquisition process. In order to regulate this conservative scoring, to guarantee the consistency of scoring, and to improve the accuracy of scoring,the benchmark scoring paper can be inserted to monitor the rating processes and errors, or a multistage rating aug- mentation pattern can be used to control the process.
出处
《心理学探新》
CSSCI
2012年第5期429-438,共10页
Psychological Exploration
关键词
主观评分
评分保守
误差控制
subjective performance rating
conservative scoring
error control
