>Bartels shows in his book that this difference is not a statistical artifact or a fluke.<p>With sample sizes of 5 and 6 (for dem/rep), Bartels shows that his results are not a statistical fluke. Sounds plausible.<p>>(It turns out that the same pattern prevails even when a Republican president is succeeded by another Republican.)<p>I take that back, sample size 2.<p>I'd be really curious to see the leave one out error for his statistics (i.e., throw away Nixon and Clinton, and see if the result persists).<p>Look, I hate Bush and McCain as much as most people (I liked Paul, however). But this book is just propaganda.