摘要
从实证的角度分析3种违背经典假设对检验的影响,得出的结论是:具有零期望独立同分布的扰动项对检验不产生影响;扰动项为异方差时,3类检验具有满意的检验效果;扰动项存在一阶序列时,负相关仍能使用传统检验统计量,正相关受到影响。
Analyzes the testing efficiency empirically from three different views under the non-classical condition.The conclusions are: if the disturbance term is independent and identical distribution with zero expectation,then the classical testing statistical is effective;if the disturbance term is heteroscedasticity,then three kinds of testing statistical can be applied;if the disturbance term is correlated with the first negative coefficient,the testing statistical also works well while it is not true for disturbance term with the first positive coefficient.
出处
《安徽工业大学学报(自然科学版)》
CAS
2007年第3期345-348,共4页
Journal of Anhui University of Technology(Natural Science)
关键词
假设检验
异方差
自相关
独立同分布
hypothesis testing
heteroscedasticity
auto-correlation
independent and identical distribution