Trying to build a responsive, dynamic webapp, that is identical in IE6 to a modern browser can easily add a lot of time. Just as an example, producing all the graphics for rounded corners, rather than being able to do them in CSS.
The thing about IE6/7 support is that it's a complete known quantity at this point. The problems & workarounds are well documented. There's tons of experienced HTML coders out there that know IE backwards and forwards. CSS frameworks like OOCSS include built-in workarounds.
So if IE support doubles the cost, it must either be something very complex, or a real nickle-and-dime project. (Either that or the webdev doesn't really what they're doing.)
That said, I'm happy to drop IE 6 where not required by the client - 8 may not have html5 or CSS 3 features, but at least it's relatively bug free.