My testing method was to focus on an object at a reasonable distance (50 feet or so) and take a picture with the 18-200mm at 200mm and the 70-200mm L at 200mm, then overlap them in Photoshop and estimate the relative cropping. It was significant. Bryan notes the same phenomenon in his review of the 18-200mm, and the pics of the starfish seem no different at 170mm vs. 200mm.
It could just be really bad focus breathing, such that anything even a little less than infinity focus results in a much shorter apparent focal length. Most people who buy a 100mm L Macro don't know they're really buying a zoom lens, and at 1:1 magnification the effective focal length is only 80mm. But in the case of the 18-200mm, I wasn't anywhere near a 'macro' distance.