请问楼主该问题解决了吗？我也想知道如何看genmod结果好坏

我只说我知道的，其他的等待大牛解答。
Goodness Of Fit是看模型fit数据的好坏，这个值看单个模型没什么意义，要不同模型之间对比才能比出好坏。如果是越小越好的criterier，后面应该还有个（smaller is better），所以lz的这个应该是越大越好。
Analysis Of Parameter Estimates就是你的模型右边，每个自变量前的参数。如果以p<0.05来衡量，tonband1这个变量所示的两个level都是non-significant的。
LR Statistics For Type 1 Analysis。SAS官网上对这个的解释是“each entry in the deviance column represents the deviance for the model containing the effect for that row and all effects preceding it in the table”。可以看出如果加了tonband1这个变量降低了模型的significance。

More words about the model:

Assumed from Poisson model (it is very likely). A large Pearson Chi-Sq may indicate an over-disperson in the data, which is one of the most typical phenomenons from count data. To remedy this problem, the model uses SCALE=PEARSON (not shown here) to account for that, such as, an over-inflated error and thus shrinking the test statistic and inflating the p value, because scaled chi-sq is one unit.
Using scale is kind of an old-fashion way in such case (as I think). People may tend to look at the over-dispersion mechanism and adopt specific measures on that, such as using different distributions.

Some comments from here.

Jingju

看AIC值