USING RESTRICTED LEAST-SQUARES TO DELINEATE THE EFFECTS OF MISSPECIFICATION IN LINEAR-MODELS

被引:1
|
作者
WALKER, E [1 ]
OBRIEN, RG [1 ]
机构
[1] UNIV TENNESSEE,DEPT STAT,KNOXVILLE,TN 37996
来源
STATISTICIAN | 1992年 / 41卷 / 04期
关键词
D O I
10.2307/2349010
中图分类号
O21 [概率论与数理统计]; C8 [统计学];
学科分类号
020208 ; 070103 ; 0714 ;
摘要
By applying standard results from restricted least squares estimation, insight can be gained about the behaviour of ordinary least squares estimates and associated tests when the restrictions imposed in the model are not true in the population. The results are applied to the special cases of under- and over-specification. The effect of misspecification is quantified on the 2(2) factorial ANOVA model involving balanced and unbalanced sample sizes.
引用
收藏
页码:467 / 476
页数:10
相关论文
共 50 条