Based on the auto- corrclation coefficient model.we gave an expression ofsize deviation of raw silk when testing- length is equal to a fraction for the unitlength,and discussed how the ratio of the size deviatinons by...Based on the auto- corrclation coefficient model.we gave an expression ofsize deviation of raw silk when testing- length is equal to a fraction for the unitlength,and discussed how the ratio of the size deviatinons by different testing-length fluctuates with the pseudo- cycle.Using these results as a basis,weexamined how the size deviation with the continuous property of raw silk sizechanges when changing the testing- length from 450 m to ll2.5m.Then weexamined the problem of quality control of raw silk.展开更多
To satisfy the demand of test evaluation of carrier rocket, the character of aerocraft and the primary flight error should be well considered. By making full use of tracking data and ground test data, a new method is ...To satisfy the demand of test evaluation of carrier rocket, the character of aerocraft and the primary flight error should be well considered. By making full use of tracking data and ground test data, a new method is provided to expand sample size through converting the system error of guidance instruments (SEGI) of special case into standard case. In terms of the prior distribution of the SEGI coefficients, a novel model, named Bayes spline model, is developed for precision evaluation, and its performance is analyzed subsequently. For a given threshold of precision,the test length can be determined by the posterior variance with the afore-mentioned model. Finally, it is proved by theoretical analysis and simulation that this method has clear engineering background and is able to fully exploit the test resources and enables us to draw a more reasonable conclusion.展开更多
测验长度(test length)是影响语言测试信度和效度的重要因素之一。本文借助概化理论(Generalizability Theory,GT)的固定侧面s×(i:p)嵌套设计和边际效用递减法则(the Law of Diminishing Marginal Utility),对中国汉语水平考试(HSK...测验长度(test length)是影响语言测试信度和效度的重要因素之一。本文借助概化理论(Generalizability Theory,GT)的固定侧面s×(i:p)嵌套设计和边际效用递减法则(the Law of Diminishing Marginal Utility),对中国汉语水平考试(HSK[中级])的测验长度进行了实证研究。研究结果显示:由130题构成的HSK[中级]测验具有相当高的测验信度,概化系数(Eρ2)可达0.8890,即使将测验的题目数量减少至120题或110题,测验的概化系数仍可以达到0.8856和0.8816(分别降低了0.38%和0.83%),这种测验长度的缩减不仅明显地降低了研发成本,而且提高了测试效率,完全能够满足标准化考试在误差控制方面的较高要求,并确保测验结果和分数解释具有较高的信度和效度。展开更多
文摘Based on the auto- corrclation coefficient model.we gave an expression ofsize deviation of raw silk when testing- length is equal to a fraction for the unitlength,and discussed how the ratio of the size deviatinons by different testing-length fluctuates with the pseudo- cycle.Using these results as a basis,weexamined how the size deviation with the continuous property of raw silk sizechanges when changing the testing- length from 450 m to ll2.5m.Then weexamined the problem of quality control of raw silk.
基金This work was supported by Special Fund for Author of Excellent Doctor's Degree Dissertation in China and the Basic Research Fund of National University of Defense Technology in China.
文摘To satisfy the demand of test evaluation of carrier rocket, the character of aerocraft and the primary flight error should be well considered. By making full use of tracking data and ground test data, a new method is provided to expand sample size through converting the system error of guidance instruments (SEGI) of special case into standard case. In terms of the prior distribution of the SEGI coefficients, a novel model, named Bayes spline model, is developed for precision evaluation, and its performance is analyzed subsequently. For a given threshold of precision,the test length can be determined by the posterior variance with the afore-mentioned model. Finally, it is proved by theoretical analysis and simulation that this method has clear engineering background and is able to fully exploit the test resources and enables us to draw a more reasonable conclusion.
文摘测验长度(test length)是影响语言测试信度和效度的重要因素之一。本文借助概化理论(Generalizability Theory,GT)的固定侧面s×(i:p)嵌套设计和边际效用递减法则(the Law of Diminishing Marginal Utility),对中国汉语水平考试(HSK[中级])的测验长度进行了实证研究。研究结果显示:由130题构成的HSK[中级]测验具有相当高的测验信度,概化系数(Eρ2)可达0.8890,即使将测验的题目数量减少至120题或110题,测验的概化系数仍可以达到0.8856和0.8816(分别降低了0.38%和0.83%),这种测验长度的缩减不仅明显地降低了研发成本,而且提高了测试效率,完全能够满足标准化考试在误差控制方面的较高要求,并确保测验结果和分数解释具有较高的信度和效度。
