Right, well the article reports 2 separate deviations that would be expected at probabilities of about .04 each -- one reported as "fewer than 4 out of a hundred" and the other reported as 4.2%. Taken together, these independent deviations are highly suggestive that the data did not occur by a chance process. The article concludes with a decidedly conservative estimate that these two deviations would be expected in about 1 out of 200 elections. My own calculation with the reported figures yields an even rarer probability of almost 1 in 600. |
|
Bookmarks