Bristol_Jonesey
Be careful.
 
I only ever normalize when I bring an audio file into Sonar that needs a bit of "beefing up" in volume.

And this.

Normalization simply discovers the highest vol in a track and raises the whole track so that point equals the normalization vol.  Something like -3 db becomes the loudest part of the track rather than, in your example, -9.7.  The rest of the track is raised in vol proportionally so you keep the dynamics, unlike compression and limiting.  It is mostly done to raise the level of a track that was recorded too low for whatever reason.  Once a track is normalized, it can be manipulated like any other audio track, tho it is likely louder.  And you don't need to "see" the waveform, of course, unless you want to edit it physically.
 
You're levels look fine and there are plenty of ways to raise the overall level of song, all of which have trade offs. Volume automation is the most natural, comp/limiters raise the average level while ... compressing the difference between the loudest and softest sounds, and normalization and physically editing peaks are destructive methods.
 
@

If hiss is audible then you are recording to low.

I think of normalization as gain changes made concrete with zero risk of overage.

A good technique is cut volume when mixing. Keep in mind as tracks are summed at the master buss there is going to be a volume increase. I like to lower all faders when I start mixing. I bring each one up as needed as I work on each track. Over time I may need to readjust faders so a track sits right in relation to other tracks. Automation is very useful here but should be applied after most all other levels are well set. Otherwise it will need to be re-automated. 
 
One shouldn't worry too much about the final volume when mixing. That should be a concern when the mix is well on its way to being finished. I like to master a stereo file of the mix not the mix itself. I don't mind going back and redo a mix if the mastering is not working. Mastering should be simple and easy. If it isn't a remix may be needed. 
 
With Sonar's Mix-recall Sonar has taken a great leap forward for being able to judge different approaches to mixing a song or project. 
 
Also try the mix as a mono mix. See if it has all its frequencies and they are in the proper relationship as the stereo mix. If you loose an instrument in the mono mix its mixed wrong.  
 
Levels when recording are not as much an issue if done at 24 bits. 16 bits will need careful monitoring to have a good level without clipping. 24 bits is very forgiving and will save a lot of hard work in the long run.
 
The above are a few tips in basic mixing.  

Don't normalize to 100% as there can still be overs from inter-sample distortion. If there are overs, even if not audible, many CD pressing plants will reject the CD due to distortion. This can also happen if the signal hits 100% for a lot of sequential samples. If someone wants MAXIMUM LEVEL they'll normalize to 0.001dB or whatever to avoid hitting 100%, but that doesn't eliminate the possibility of inter-sample distortion.
 
BTW I love getting tracks to master that have 6 to 12 dB or headroom, gives me a bigger sandbox to play in.
 
 

Anderton
Don't normalize to 100% as there can still be overs from inter-sample distortion. If there are overs, even if not audible, many CD pressing plants will reject the CD due to distortion. This can also happen if the signal hits 100% for a lot of sequential samples. If someone wants MAXIMUM LEVEL they'll normalize to 0.001dB or whatever to avoid hitting 100%, but that doesn't eliminate the possibility of inter-sample distortion.
 
BTW I love getting tracks to master that have 6 to 12 dB or headroom, gives me a bigger sandbox to play in.
 
 

meaning that your pre-master peaks are usually between the -6db and -12db range?  Do you use K-metering at all?

alewgro
 
meaning that your pre-master peaks are usually between the -6db and -12db range?

 

Do you use K-metering at all?

 
Sort of...it's a variation I came up with before K-metering was around. But I should really switch over to be in sync with the rest of the world. There's always a first time 
 

Anderton

alewgro
 
meaning that your pre-master peaks are usually between the -6db and -12db range?

 

Do you use K-metering at all?

 
Sort of...it's a variation I came up with before K-metering was around. But I should really switch over to be in sync with the rest of the world. There's always a first time 
 

yea i just moved over, but its weird... they say to master at K-12, but K-12 seems to be about 3db lower average RMS than typical modern day produced.   I'm guessing people are mastering around K-10 or K-8 now but most meters dont go that high 

Do the numbers in your post indicate a 20 db difference between the average signal and a solitary peak?  If so, that is not bad, but it is a pretty large dynamic range for pop music.  If that is the case, you might see if there is just one spot causing that -6db peak, in which case maybe you've got a tom that needs a high-pass, or some other one-time anomaly.  The only reason that matters is that you don't want to have to master the whole song at a low level just to account for one peak.
 
If that's not the case, then ignore my comments. :-)  A -6db peak is fine for a track to be sent off for mastering (although I am guilty of usually cutting hotter).

alewgro
yea i just moved over, but its weird... they say to master at K-12, but K-12 seems to be about 3db lower average RMS than typical modern day produced.   I'm guessing people are mastering around K-10 or K-8 now but most meters dont go that high 

 
That's so interesting to me. I mostly listen to stuff I'm mixing and that is almost always to -14. I strain to get to -12, putting up resistance.