邹检验不能检验出结构变化的形式，建议使用虚拟变量模型．然后检验虚拟变量前的系数是否显著的大于０．

李子奈的《计量经济学》第二版写的很好，还有事例．

来自EViews的Help文件

As an example, suppose we estimate a consumption function using quarterly data from 1947:1 to 1994:4 and specify 1973:1 as the first observation in the forecast period. The test reestimates the equation for the period 1947:1 to 1972:4, and uses the result to compute the prediction errors for the remaining quarters, and reports the following results:

Neither of the forecast test statistics reject the null hypothesis of no structural change in the consumption function before and after 1973:1.

If we test the same hypothesis using the Chow breakpoint test, the result is:

Note that both of the breakpoint test statistics decisively reject the hypothesis from above. This example illustrates the possibility that the two Chow tests may yield conflicting results.

你可以找这本书＜数据分析与Ｅviews应用＞，第４６面，对于你的问题，阐明的十分详细且易懂

易丹辉主编，中国统计出版社，２００２年