【作者】 陈德枝；

【导师】 戴海琦；

【作者基本信息】 江西师范大学 ， 基础心理学， 2009， 博士

【摘要】 动态评估是一种关注个体未来发展水平的互动式测量模式。它通过教学和干预等把个体的学习过程和学习结果结合起来,考察个体的未来发展水平或学习潜能。与静态测验相比,动态评估可以提供更多的信息,如有关学习过程的信息等(Brandsford等,1987)。动态评估的理论基础主要来自皮亚杰的认知发生论和维果茨基的最近发展区学说等,评估对象往往是特殊需要儿童和非主流文化儿童。传统的动态评估常用前、后测分数的差值来量化学习潜能的大小,而无法对两极端获益分数进行探讨;尤其是当前后测验不平行,项目难度存在差异时,更难以对评估结果进行解释;另外,由于经典测验理论对分数的解释基于样本分布,很难适应动态评估。为了解决这些问题,Embretson等提出了用IRT模型作为动态评估的计量学模型。虽然基于IRT理论的动态评估克服了传统动态评估的许多缺陷,但它还是有着与后者一样的问题:即最终报告的是前、后测的能力改变值,无法对能力变化背后的心理加工过程进行探讨。而认知诊断将认知心理学和心理测量学结合起来,根据任务的心理认知加工过程来设计、编制测量工具,分析和解释测量结果。因此,结合认知诊断和动态评估的思想和方法,可以设计与开发基于认知诊断的动态评估方案,并从宏观与微观两个层面同时对方案的实施结果进行分析与研究。本研究将认知诊断模型MLTM for MS和动态评估IRT模型MRMLC作为认知诊断和动态评估相结合的切入点,结合一般小学儿童图形推理认知发展的特点,开发和设计了一套基于认知诊断的小学儿章图形推理能力的动态评估方案。方案采用的动态评估模式为:前测—干预1—后测1—干预2—后测2。整个过程共有两次干预和三次测量,每连续两次测量之间间隔1～2周,每次后测前均进行一次标准化的教学干预(干预1和干预2),每次干预都是根据儿童图形推理认知发展的特点而设计的。围绕本方案的设计,展开了一系列的研究,包括测验的编制,方案的实施、结果的分析以及评估结果的质量分析等。具体如下:首先,基于儿童图形推理认知发展的特点,提出编制动态评估测验的基本原则,包括符合评估目的、儿童发展特点及数据结构设计等五项原则。然后基于这些基本原则和图形推理项目的认知分析,编制了用于评估的三套标准图形推理测验。为检验这三套测验是否符合数据结构的基本原则,在江西省北部选取没有作答过图形推理测验的230名二到六年级的农村小学儿童,同时施测这三套测验。CFA分析结果表明,这三套测验在结构上是相等的,符合测验编制的基本原则。然后,结合儿童图形推理认知发展特点和MLTM for MS,对这三套测验编制了相应的认知成分测验。其次,根据MRMLC模型的特点,设计了两种能力参数估计的方法:一次性估计法和分别估计法。这两种方法的计算机模拟比较研究结果表明,一次性估计法的准确性和稳定性要稍好些。因此,在随后的分析中均采用该方法。接着,在江西省北部选取两所农村小学四到六年级共182名儿童,参与本方案的实施。结合MRMLC和MLTM for MS,对儿童动态评估结果进行宏观和微观分析与研究。宏观分析与研究结果表明:(1)各个年级在后测2所测到的学习潜能与0相比均有显著性差异(p＜0.05),而且,在后测2所测到的学习潜能都显著地高于在后测1所测到的学习潜能(p＜0.05);(2)在标准图形推理能力的当前水平上,五年级与四年级之间没有显著性差异(p＞0.05);但六年级的当前水平极显著地高于四年级和五年级的(p＜0.01);(3)三个年级两两之间在总的学习潜能上没有显著性差异(p＞0.05)。微观分析与研究结果表明:(1)通过结合MRMLC和MLTM for MS,可以实现精细地分析与比较每位儿章在图形推理的各认知成分上的变化;(2)认知成分动态评估结果的年级间比较表明,在当前水平方面,五年级比四年级更擅长使用规则应用成分;六年级又比五、四年级更擅长运用规则构建和规则应用成分;在学习潜能方面,四年级比五、六年级能够更擅长使用视知觉识别和规则应用,但六年级比四年级能够更擅长运用规则构建成分;(3)各年级认知策略的动态评估结果表明,四年级能够更正确地运用知觉分析策略解答项目;五、六年级能够更正确地运用分析策略来解答项目。最后,从信度和效度两个方面对该方案的实施结果进行质量分析。研究结果表明,每套标准图形推理测验的内部一致性均较高;本方案对儿童图形推理学习潜能的评估是有效的。更多还原

【Abstract】 Dynamic assessment is an interactive measurement model which puts emphasis on the individual psychological development in the future.Dynamic assessment,which combines individual’s leaning process and learning results through teaching and intervention,can take both individual’s future level and learning potential into consideration.Compared with static testing,dynamic assessment can provide more information,such as the process of learning(Brandsford.etc,1987).With Piaget’s cognitive theory and the concept of the zone of proximal development(ZPD) from Vygotsky as theoretical base,the dynamic assessment take disable children and sub-culture children as its subjects.The traditional dynamic assessment quantifies the learning potential by comparing the difference of scores between the pre-test and the post-test,which fails to explain the individuals who have extreme scores.It is a more challenging work when the pretest and the posttest are not parallel test.In addition,as the CTT is based on the distribution of the sample,it is unsuitable for dynamic assessment.Consequently,Embretson.etd.put forward the psychometric model based on IRT to solve these problems.Although the dynamic assessment based IRT has overcome many shortcomings of the traditional dynamic assessment,it still has the same problem as the latter,that is,the final report of the measurement is the modifiability between the pretest and the post-test instead of the psychological processes underlying the ability to change.Cognitive Diagnosis,however,according to the psychological processes,designs measurement tools and interprets the results of measurement by combining cognitive psychology and psychometrics.Therefore,the design of dynamic assessment based on the cognitive diagnosis by the combination of the ideas and methods of cognitive diagnosis and dynamic assessment can be developed to analyze the results of the program from both macro-and micro- level.The study,which initiated from cognitive diagnosis MLTM for MS and MRMLC of the dynamic assessment model,developed a dynamic assessment design based on cognitive diagnosis to measure the learning potential of the figure reasoning of the elementary school children according to the characteristics of the cognitive development of figure reasoning of the general elementary school children.The model of the dynamic assessment in the study is:pretest-intervention 1-posttest 1-intervention 2-posttest 2.There are two measurements and three interventions during the whole process,with 1-2 week interval between every two consecutive measurements.Prior to each post-test,a standard teaching intervention based on the characteristics of figure reasoning of the elementary school children was performed.Meanwhile,a series of studies on the design have been done,including compiling of the items,performing of the design,analysis of the results and even the quality of the results.They are listed in the following:First,according to the cognitive development of the figure reasoning of the elementary school children,the author put forward five principles of compiling the test, including the objective of the assessment,the data structure design and etc.Next,three sets of the standard figure reasoning tests were compiled after the cognitive analysis of these principles and figure reasoning.Then,230 elementary school students ranging from Grade Two to Grade Six,all of whom have not any experience in answering the item of the figure reasoning from the rural elementary schools in the north part of Jiangxi province were chosen as subjects to verify whether the three sets of tests are in line with the basic principles of data structure.The results of CFA analysis show that the three sets of tests are equivalent in structure and up to the basic principles.Finally,the author compiled the cognitive tests for each set of the tests based on the MLTM for MS based on the cognitive development of the figure reasoning of the elementary school children.Secondly,according to the characteristics of MRMLC,the author designed two methods for estimating the parameter of ability.They are one-off estimation and respective estimation.The results of the comparative study of computer simulation show that,the accuracy and the stability of the one-off estimation are better than the respective estimation method.Therefore,the former one was used to estimate the ability in the subsequent study.Next,182 children ranging from grade four to grade six coming from the rural elementary schools in the northern part of Jiangxi Province were chosen as sample to participate the study of the implementation of the design.The author analyzed the results of the implementation from marco- and micro- level after combining MRMLC with MLTM for MS.The results of the Macro-analysis show that:(1)there is significant difference between zero and the learning potential measured by the posttest 2(p＜0.05), and the learning potential measured by the posttest 2 is is significantly higher than that measured by the posttestl(p＜0.05);(2) on the current level of the standard figure reasoning test,there is significant difference among the students from the fifth grade and fourth grade(p＜0.05);(3) on the total learning potential,there are no significant differences(p＞0.05).The results of the Micro-analysis show that:(1) the design in this study,by combining MRMLC with MLTM for MS,can measure individual’s change during the cognitive process of figure reasoning successfully.(2)The comparisons of the cognitive component among the three grades show that,on the current level of figure reasoning, the fifth grade students are more accomplished in using rule application than the fourth grade students;while the subjects from the sixth grade are more accomplished in using the rule construction and rule application than the fifth and fourth grade student;on the total learning potential,the students of the fourth grade can become more accomplished in using the visual perceptual cognition and rule application than the fifth and sixth grade,while the students from the sixth grade can be more accomplished in using rule construction;(3)as to the cognitive process strategies,the learning potential of the fourth grade students manifest mainly in the fact that they can use the perceptual analytic strategy to answer the item in a more accurate way;while the fifth and sixth grade students do better in using the analytic strategy.Finally,both the reliability and validity of the results of the implementation of the design have been tested.The results show that the internal consistency of every standard test is high and the learning potential of figure reasoning assessed by the design is valid.更多还原