对学前儿童语言学习能力诊断量表的效度评价

A Validity Study on the Diagnostic Tool for Language Learning Ability of Preschool Children

以所编制的量表为工具 ,对采集的数据进行效度分析 ,结果表明各分测验与全量表有较好的相关 ,说明量表的内容效度是比较高的。使用因素分析的方法 ,将全部变量作系统分类 ,研究量表的结构 ,绝大部分分测验在所得的四个因素上的共通性都大于 0 .70 ;保留下的分测验与所属因素的相关系数在0 .5 3 -0 .84之间 ,它们在各个因素上有较高的负荷量 ,说明量表有较好的结构效度。从效度分析的结果看 ,本量表的测量结果应该是准确的。另外 ,还根据因素分析结果指示的方向 ,调整了分测验 ,调整后的量表结构不但与假设的量表结构十分吻合 ,而且更条理化。

Language learning difficulties have become a major concern in the field of special education in China. Because the majority of measures designed to assess language learning difficulties for English may not be applicable to Chinese language, it is critical that new measures be developed for Chinese. This will permit accurate identification of children with language learning problems. A theoretical framework was developed by combining psycho linguistic disciplines and the nature of Chinese language acquisition (Ref. 3) . Using this framework, a measure titled 'A diagnostic tool for language learning ability among preschool children' was developed for diagnosing language learning ability of children aged 3 6. The measure contained 10 sub tests with 255 items. The 10 sub tests were visual discrimination, audio speech response, visual motor response, audio memory, visual memory, nonverbal expression, and verbal expression. A sample of 210 children aged 3 6 was randomly selected from urban districts of Beijing. Data collected from testing of these children was assessed in a study designed to measure the reliability and validity of the measure. The collected data were entered into a database FOXBASE software and then analyzed using SAS. The sub tests of the measure will be referred to as 'A' through 'J'. The first step in the data analysis was to evaluate correlation coefficients between scores of items and of sub tests as well as between scores of items and of total measure. Because the sub test 'J' consisted of 4 testing contents, the original 10 sub tests were treated as 13 sub tests. The result of this analysis was applied to select 143 items from 255 which showed positive and favorable correlation. For the selected items and the sub tests, a second set of correlation coefficients were computed in order to study content validity. The results showed that a majority of sub tests had high correlations with the whole measure (the correlation coefficient was higher than 0.61). The variable measure was analyzed by factor analysis to describe the construct of this measure. The factor analysis and rotation showed that all sub tests contributed to four factors and their community were higher than .70. This analysis suggests high construct validity. A positive relationship between test scores and age was also obtained. In addition, a normal distribution was found by grouping each level as 10 scores, from the lowest (20) to the highest (140). The results of the above analysis showed that favorable correlation existed between the sub tests and the measure as a whole test. These results indicate that the measure could be applied to test children's ability in language learning. The results of data analysis was also applied to revise the measure. The process of this data analysis accepted the hypotheses made for constructing the measure. Due to deficiencies of sub tests 'E' and 'F', these two sub tests were deleted from the measure. Finally, the total measure contained 9 sub tests and contributed to four factors. According to the nature of Chinese language acquisition, factors were named as visual stimulus, audio stimulus, visual audio memory, and visual-audio discrimination. In conclusion, this study used correlation techniques and factor analysis to examine content validity and construct validity of a measure. The results of data analysis supported the construction of the measure. The whole analysis procedures were also very useful for revising the measure.