元分析中统计异质性的检验:一项Monte Carlo研究

元分析中,异质性检验常为模型的选择和调节分析提供重要的参考依据.有多种方法和途径对异质性进行评价,包括Q,H,I2和似然比检验.通过模拟研究比较,发现单一的Q统计量可能导致错误的模型选择,H和I2值没有因为估计方法(矩估计和似然估计)的不同而对异质性的检验结果存有较大差异,是一个有效可靠的辅助统计量;Q统计量与似然比检验相比,在I类错误率上控制较严格,但统计功效不如似然比检验.研究结果表明,可以选用Q统计量结合报告H或I2值的方式作为异质性检验的最佳方案.     

工作记忆中汉语词长效应机制研究

工作记忆中的词长效应是指在即时序列回忆中短单词词表的成绩优于长词表的现象.实验一对视听感觉通道回忆成绩进行比较,发现听觉呈现下字词回忆率将要高于视觉呈现,且两种感觉通道下都会有明显的词长效应;不同回忆方式均产生词长效应,肯定了词长效应中识记阶段的复述过程的作用.实验二在视觉呈现下的即时回忆中加入发音抑制,汉字词长效应消失,这正是语音回路理论对词长效应的理论预期.通过项目错误分析,比较有无发音抑制下长短词的三类错误,结果进一步支持了语音回路理论对词长效应的解释.

Abstract：

Word length effect in working memory means that shorter items are remembered better than longer items in immediate serial recall tests. In the first experiment reported in this paper, the recall performance under visual and auditory modalities was compared and it was found that the result of aural presentation was better than the result of visual presentation and word length effect was detected in both. Standard word length effects were noticed in serial recall and free recall tasks. These results supported that word length effect is produced not only in rehearsal process of the memory stage, but also in pick-up process of the recall stage. In a second experiment, articulatory suppression was added under the conditions of visual presentation in immediate recall, and word length effect was found to disappear, as proposed by the phonological loop hypothesis. In addition, we also compared three kinds of mistakes for no articulatory suppression and articulatory suppression conditions. The list of short words was more influenced by articulatory suppression. The results supported the explanation of phonological loop theory for word length effect.