After the 2004 election I had this idea that Bush’s victory over Kerry is analogous to Truman’s over Dewey in 1948. In both cases, the incumbent had attained office indirectly (through Roosevelt’s death for Truman, and with a popular vote minority for Bush), had not-so-great approval ratings during the campaign, and was expected to lose to a challenger who looked strong on paper. But again, in both cases, the incumbent ran a combative campaign appealing to his base and surprised many with a strong victory for himself and his party. But then after all was over, the incumbent’s popularity remained fairly low.

I don’t really know what to make of this reasoning. My perceptions of Truman and Bush are both mostly from written sources. Obviously the two men have a lot of differences, but I do remember that G. W. Bush was perceived as some sort of centrist in the 1990s and now seems more of a strong conservative, and similarly Truman seemed to have moved, at least in perception, from a centrism to strong liberalism.

A key issue in the analogy, I suppose, is the implicit suggestion that Bush in 2005-2008, like Truman in 1949-1952, could see a continuing decline in approval followed by a loss in his party’s control of Congress. Or, to put a different spin on it, that in retrospect Truman is generally viewed positively, in that some of his domestic and foreign policies that were unpopular at the time but are now respected as principled.

I don’t really know how to think about this sort of analogy. I guess the key connection is the election win without a strong approval rating. That Bush and Truman both have the image of fighting partisans (e.g., the famous poll that said that half the people thought that Bush was a “uniter not a divider” and half didn’t) also seems relevant to me. Perhaps this idea could be studied statistically by looking at the relation between Presidents’ approval ratings and their abilities to achieve policy goals. After controlling for party control of the houses of Congress, how predictive is Presidential approval of Presidential policy successes? I imagine someone has studied this.

Some barely related comments on statistical graphics

P.S. Here’s a paper by Charles Franklin on graphing Presidential approval (or other time series, for that matter). I like the paper–its Figure 5 has a particularly clear plot of all the approval series on a common scale. (Clinton and Roosevelt are the only Presidents with generally upward trends in approval–I didn’t know that.)

Franklin’s Figure 5 has a lot of little features that make it more useful and readable:

– Most important, the plot for each of the 12 presidents (from Roosevelt to G. W. Bush) is small (but clear), so there is room for all 12 to fit together on a small page.
– There are light guidelines at 1 year, 2 years, 3 years, and 4 years, so the reader can easily see when different things are happening.
– The approval series themselves are shown by thin lines so that the time spacing of the data is clear. I can’t tell you how many graphs I’ve seen that were marred by lines that were too fat.
– The y-axis goes from 0 to 100. This makes sense since these are the theoretically highest and lowest approval ratings. Little would be gained by restricting to the range of the data (approximately 20% to 95%). This is obvious, but graphics programs will restrict by default, so Franklin gets credit for keeping the full scale in for clarity in interpretation.

I also have some minor criticisms of the graphical presentation:

– The 12 presidents are listed in a 3×4 grid, with Roosevelt at the bottom left and Bush at the upper right. Given how we usually read, I’d prefer to start with Roosevelt on the upper left and then go across and down from there.
– The time scale is presented in months. Years would be better. It’s not so important since the guidelines at each year are given, but still, labels at 0,2,4,6,8 would be best.
– The presidents are listed by two-to-four-letter abbreviations. Some are clear (FDR, JFK), some required puzzling out (HST, GRF, JEC). But why bother abbreviating? There’s enough space in the label area of each plot to write “Roosevelt”, “Truman”, etc.

More . . .

P.P.S. Searching “bush truman” on the web, I found this article by William Stuntz making the same analogy–and he gets extra credit, having made the connection in October, 2004. I like Stuntz’s article, but I was disappointed to see that he made the usual “red-blue error”: he writes, “it’s the Democrats who are strongest among upscale voters in New York and Boston, and the Republicans whose base spreads across the South and West.” He’s right about the geography but wrong about the “upscale.”

There’s also this connection of Bush and Clinton by Tim Hames from 2003, and this by Saul Singer in 2002, which focus on the idea that both Presidents were underestimated. (And here, here, and here are some links from the other side of the political spectrum.)

I also see that Larry Sabato drew the Bush-Truman analogy in May, 2004, but without taking the next step and considering what could happen in the years after an unpopular president is reelected.

Finally, see here for lots more if you’re into this sort of thing. Depending on how it’s presented, the Bush-Truman analogy can be flattering to Bush, flattering to Truman, or unflattering to either. The pattern is there, but it can be interpreted in different ways depending on how you feel about the two presidents.

Also here and here are popularity comparisons of Bush and Nixon, but I still think Truman’s a better comparison because it is gradual rather than the result of a single scandal.

P.S. Still more here on the topic.