问题：

  各位老师好：

我是重庆大学博士研究生张宽，我最近收到一个论文外审专家对计量建模的建议，其中提出了Madona effect，并给出了参考书目，自己也下载了该书，但是还是无法很透彻理解这个问题，特向各位老师请教。其中专家意见如下：

- However, for avoiding the Madona effect in statistics manipulation (acc. to Georgescu-Roegen, Nicholas, The entropy law and the economic process, Harvard University Press, 1971) statistics my proposal is to use 2-3 methods in parallel, sustained by the literal explanations.

书中关于Madona effect 描述如下：

Besides, if one formula does not pass the test, we can always add another variable, deflate by another, and so on. By cleverly choosing one’s chisels, one can always prove that inside any log there is a beautiful Madonna.（340页）

在附近中上传了该书。

答：

ivpoisson cfunction patents_SQ i.year i.cic2 (r5lngdp  =road ),vce(jacknife)。迭代不停止，试试这个。

Besides,if one formula does not pass the test, we can always add another variable, deflate by another, and so on. By cleverly choosing one’s chisels, one can always prove that inside any log there is a beautiful Madonna.这段读懂就理解了，这个估计不是主流说法。请参照下面：