MODIFIED WALD TESTS IN TESTS OF EQUALITY BETWEEN SETS OF COEFFICIENTS IN TWO LINEAR REGRESSIONS UNDER HETEROSCEDASTICITY
MODIFIED WALD TESTS IN TESTS OF EQUALITY BETWEEN SETS OF COEFFICIENTS IN TWO LINEAR REGRESSIONS UNDER HETEROSCEDASTICITY
复制标题
异方差下两个线性回归系数组间等式检验的改进Wald检验
DOI：
10.1111/j.1467-9957.1986.tb01266.x
复制
发表时间：
1986
期刊：
影响因子：
--
通讯作者：
Hazuhiro Ohtani
中科院分区：
文献类型：
--
作者：
Yuzo Honda;Hazuhiro Ohtani
当我们想检验两个线性回归中系数集之间的相等性时，我们使用Chow（1960）的检验。这些检验假设两组观测值之间的干扰方差相等。丰田章男（1974年）;施密特和西克尔斯（1977年）以及大谷和丰田（1985年）发现，上述测试之一（两个样本都有足够数量的。当两个样本的干扰方差实际上是异方差时，观测）可能具有相当差的性能。Jayatissa（1977）和Watt（1979）提出了两种替代检验，这两种检验都没有假设两个样本之间的干扰方差相等。Jayatissa的测试具有作为精确测试的优点。但它同时也有一些缺点。1 Watt提出的检验是Wald检验，具有最优渐近性质。然而，人们对其小样本行为知之甚少。Monte Cario在Wattand in本田（1982）的实验表明，当两个样本大小为中等或更大时，Wald检验优于Jayatissa检验。但是，当样本量很小时，无法得出确切的结论。Wald检验的问题在于
When we want to test equality between sets of coefficients in two linear regressions, we use the tests by Chow (1960). These tests assume equality of disturbance variances between two sets of observations. Toyoda (1974); Schmidt and Sickles (1977); and Ohtani and Toyoda (1985) found that one of the above tests (the test in the case where both samples have sufficient numbers of. observations) may have a rather poor performance when disturbance variances of two samples are in fact heteroscedastic. Jayatissa (1977) and Watt (1979) proposed two alternative tests, neither of which assumes equality of disturbance variances between two samples. The test by Jayatissa has the advantage of being an exact test. But it has some disadvantages at the same time. 1 The test proposed by Watt is the Wald test and has the optimal asymptotic properties. However, little is known about its small sample behaviour. Monte Cario experiments inwattand in Honda (1982) revealed that the Wald test is preferable to the Jayatissa test when two sample sizes are moderate or larger. But no firm conclusions can be drawn when sample sizes are small. The problem of the Wald test is that the