It's time for some numerical politcal fun
that doesn't involve polling!. As hard as it is to believe, there is more to life than margins of error and subgroups.
Below, I'll take a look at the reading level and distribution of word and sentence length of what Bush and Kerry had to say on Thursday, why Kerry came across more effectively (aside from the visuals and stuttering, which were very obvious), and why there's a very good chance John Edwards will lose the VP debate.
The Flesch-Kincaid reading level model is the number that Microsoft Word spits out if you have it grammar check your essay. For a point of reference, the claim is that "standard writing" is written at an 8th grade reading level; the New York Times is around 10th grade or 11th grade.
This article in the Globe and Mail (subscription required) shows only a small difference between the reading level of Bush's speech and that of Kerry's speech; Kerry is below an 8th grade level, which is the first time the Democratic candidate has been so child-like is his speech in a long time. Kerry clearly had a mission to speak directly and avoid the "ponderous, boring" tag.
But does this tell the whole story? No. Reading level formulas are based on the average word length and the average sentence length. But as well all know from discussions of Bush's tax scheme, the average doesn't tell the whole story. Afterall, do these two passages sound like the same speaker?
'I know I can do a better job in Iraq. I have a plan to have a summit with all of the allies, something this president has not yet achieved, not yet been able to do to bring people to the table.'
'My opponent says help is on the way, but what kind of message does it say to our troops in harm's way, "wrong war, wrong place, wrong time"? Not a message a commander in chief gives, or this is a "great diversion."'
Each passage is two sentences; well, each is transcribed as two sentences. Almost the same number of words, almost the same number of letters. The Flesch-Kincaid method (and other readability scoring systems) would say they are roughly the same. But who do you want to be your president?
Examining the distribution of word and sentence length makes this even clearer.
Metric Bush Kerry
Word Length 4.30 4.28
% words, 2-3 letters 24.2 22.5
% words, 10+ letters 6.2 6.8
Sentence Length 12.3 14.4
% sentences, 1-15 words 70.8 64.1
% sentences, 25+ words 8.63 14.9
See the difference? Kerry won the debate by carefully oscillating between speaking directly and elaborating. Bush on the other hand, can only speek in stacatto phrases. It's amazing what four years of speaking only in sound bites will do to you, eh?
But enough with the past; all elections are about the future. There are a number of differences between the presidential debate and the VP debates. The audience for the VP debate is usually more female, more undecided, and has more informed voters. Not surprisingly, the debate often covers more details, and is less "dumbed down". Here's a graph showing the reading level of the Gore-Lieberman debate as compared to (search for "figure 3" to get to the right chart). Now Lieberman and Cheney are both wonks, so maybe there was an amplifying effect. But Cheney is clearly more schooled in policy details than the President is. Edwards, on the other hand, has spent 6 years in public office, and spent 3 of those running for President. I suspect he can hold his own on local issues (agriculture), or on areas of interest to him (medical issues, particularly PBoR). If we get into a debate on trade or taxation or who knows what else, Cheney could very easily out-detail Edwards and project gravitas quite well.
[For those wondering why the number is so low in general, remember, there isn't necessarily much gain to speaking at a high reading level. After all, if your sentences contain so many subordinate clauses that everyone listening gets lost, what have you accomplished?]