Whew, there's several ways to skin this cat.
Me personally, I would move the different takes to their own tracks so I could use the gain staging to get them in the ballpark to each other. It makes SONAR's native tools that much more powerful since there's a few different places to adjust these levels.
Alternately, you could use Clip Gain envelopes on each clip to help get these vocals roughly in the same volume zone. I do this all the time with full vocal takes of a song, and by doing so you don't have to compress as much, and can use the compressor more for character than for correction.
You COULD use a volume envelope over the entire track, but that might get a bit laborious, and I think that envelope is post-fx (double check me on that).
Of course, you could just compress the entire track, but I wouldn't. Finally you could do it destructively, and use the Process|Gain or Process|Normalize functions to "hard code" your audio to do these edits.