中国高校英语专业写作测试效度检验研究
2010-12
上海外语教育出版社
修旭东
361
本书由修旭东所著，从受试者特征、理论效度、内容效度、评分效度、试后效度和效标关联效度方面对我国高校英语专业八级测试写作项目和(英语专业)毕业论文评价的效度进行了检验。其参与者来自中国三所大学的学生和中国及英国多所大学或学院的专家和老师，研究工具包括：有声思维实验、问卷、个人访谈、小组访谈和日志。研究表明，尽管不同类型的效度证据表明两者的效度尚可，但仍有改进和提升的空间，尤其是在内容效度、评分效度和试后效度方面。
INTRODUCTION
PART I LITERATURE REVIEW
CHAPTER 1 LITERATURE REVIEW I-LANGUAGE TESTING AND VALIDITY
EVIDENCE
CHAPTER 2 LITERATURE REVIEW II-PARAMETERS FOR A VALIDATION
FRAMEWORK FOR AWT
PART II PILOT STUDIES
CHAPTER 3 THE DEVELOPMENT AND PILOTING OF INSTRUMENTS AND
PROCEDURES FOR APPLYING THE VALIDATION FRAMEWORK TO TEM8E AND
GRADUATION DISSERTATIONS
PART III MAIN STUDY
CHAPTER 4 DISCUSSION OF THE RESULTS CONCERNED WITH THE EVIDENCE OF
TEST-TAKER CHARACTERISTICS
CHAPTER 5 DISCUSSION OF THE RESULTS CONCERNED WITH THEORY-BASED
VALIDITY DEIDENCE
CHAPTER 6 DISCUSSION OF THE RESULTS CONCERNED WITH CONTEXT
VALIDITY EVIDENCE
CHAPTER 7 DISCUSSION OF THE RESULTS CONCERNED WITH SCORING
VALIDITY-,CONSEQUENTIAL VALIDITY-,AND CRITERION-RELATED VALIDITY
EVIDENCE
CHAPTER 8 CONCLUSIONS AND IMPLICATIONS
REFERENCES
APPENDICES
some ofthe responses that can be scaled. Latent variables refer to those （constructs）theoretically hypothesized, or those underlying ones that can not be directlyobserved or measured. In SEM, these latent variables consist of exogenouslatent variables and endogenous latent variables. Exogenous latent variablesrefer to those that can affect the other latent variables but also be affected bythe observed variables outside of the model, Endogenous latent variables referto those that can be affected by both exogenous latent variables and observedvariables. Those observed variables that affect the exogenous latent variablescan be called exogenous indicators and those observed variables that can affectendogenous latent variables are called endogenous indicators. As SEM is usually used for confirmatory research, before conductingthe modelling procedures, we first have to know what the already existenttheoretical framework or model is to test or confirm and what are therelationships between observed variables and latent variables and betweenlatent variables themselves in the framework. Although it is usually usedfor confirmatory research, exploratory research is also conducted in orderto explore whether the relationships are correctly established. However, onthe other hand, if too many observed and latent variables are involved in themodelling, the model can not be established. In our present study, we initially tried to establish all the relationshipsbetween the observed variables we have designed in all questionnaires andthe latent variables extracted from the test-taker characteristics, and sub-processes of academic writing, but failed because too many observed variablesand latent variables can not be sorted out by present LISREL （8.50）. We finallydetermined only to seek the relationships between: 1） academic writingstrategies （targeted by QA2, especially the second section, on which theresults of regression analysis and ANOVA displayed the highest percentageaccounting for the variance of the score gains）, 2） sub-processes （reflectedby QB1）, and 3） gains in TEM8 （which consists of WRTN score and totalscore）. ……
质量不错，只是还未看内容。