As Bruce and Scrubs mentioned, the reason why you feel normalization is probably sounding better is that possibly you are not using gain makeup in the compressor, but just cutting down on the dynamic range.
The pupose of the compressor (at least the usual purpose) is to reduce the dynamic range by lowering the level of the audio by a set ratio and threshold. By doing that you can raise the overall volume to create a track with an overall louder average (RMS). This is where you need "gain makeup" to makeup the difference between the original signal and the compressed signal.
With normalization (in the traditional sense) the software goes through the entire track finding the loudest value. If it's less than 0 dbFS (digital peak) it will raise the volume of the audio so that the entire track is raised proportionally to get as close to the peak as you tell it to. This does not decrease the dynamic range, but as Bruce mentioned just raised the volume of the track by the preset amount. If later you decide the vocal are too loud and reduced the volume you acomplished nothing other than to add distortion to your track.
Now if you're saying that you don't like the sound of compresion, that's an entirely different topic I'll need my third cup of coffee of the day for ...