主题：【原创】两党选战与经济问题 -- dfindy

碰巧纽约时报有一片与此观点相同的观点文章

http://www.nytimes.com/interactive/2008/10/14/opinion/20081014_OPCHART.html

我找到了一个网上反驳的文章，主要意思是这个结论不符合统计方法的。原文是味了结论而找的数据：

All right, here is the solid explanation step by step:

1. First, to qualify as a "statistic", it has to have an argument and use

data to support/overturn that argument. It does not. But that's fine, since

we know what it wants to say.

2. The purpose of the article is to support the argument (hypothesis) "stock

market fares better under dem prez".

3. That translates to "average returns of stock market under Dem is greater

than that of under Rep"

4. It provides 13 data points. One of them is clearly an out-lier. Let's

ignore that for now.

5. The two groups, Reps and Dems, have mean/variance of 4.84%/0.0056 and 8.

67%/0.0011. Don't ask me why these numbers differ from those given in the

article. I candidly do not know.

6. The t-statistics of the difference of the two means is therefore

calculated to be 0.61

7. With 12 data points in total, this translates to a 'P-value' of 0.55

8. In statistics, we need a P-value less than 0.05 to make a "sensible"

argument (i.e., the result is "statistically significant"). So, draw your

conclusion here.

9. How about the Hoover out-lier? With that data point, the P-value is 0.39.

10. Without telling you all these, this article implies a wrong conclusion.

Therefore, it is not a statistic based on scientific procedure.