I'm still a novice when it comes to mixing/mastering, and am a bit confused.
 
It seems like some folks like a simple single band compressor at the mix output like the SSL buss compressor.
 
It also seems that some folks like to apply multiband compression like the LP multiband from Cakewalk.
 
I'm sorry if this question is too vague, but why would one choose one over the other?  Do you use both?
 
Any advice would be appreciated.

I'm no expert either, but it appears to me that single band mix bus compression is desirable for the "gluing" effect, whereas multiband compression is commonly used as a tool to fix a problem with the mix, something that could theoretically be dealt with by going back and fixing it on the track level. I don't think of mix glue as a "fix," rather something which adds cohesion to the mix that you couldn't get on the track level. Having said that,  I have used a multiband compressor on single tracks before, particularly to control unwieldy low end on synth bass. 

Some time ago I read an interesting discussion about compression in the Cockos forum. I cannot post links yet so google for "Compression??? I just don't hear it!!". The most useful comment I found there was from the member yhertogh, giving a formula to adjust compressors that he found in the book "Mixing with your mind" by Mike Stavrou. There are good references on the web and I personally recommend the book "Mixing Secrets for the Small Studio" by Mike Senior.
 
Multiband compressors are more useful operating over full band signals where some parts of the spectrum are more prominent than others. You can set the frequency bands accordingly and adjust the parameters for each one independently so, for instance, a strong bass would not trigger the compressor's gain reduction excessively and attenuate (or mask) other regions of the spectrum. The typical use is in mastering, were your source is the entire mix. Operating on individual channels you normally don't have this problem, as you naturally need to restrict each channel's bandwidth with equalization to avoid clashing.
 
A slight full band compression applied to the mix bus is a particular case where the goal is to achieve a better "blend" of all the signals arriving there. In this case the fact that all the signals are interfering with each other on the compressor helps to achieve this (note that the key word here is slight).

http://forum.cockos.com/showthread.php?t=57648

You can experiment with multi band compression....but IMHO, if you are a novice or beginner at mastering, you should stick with simple Compression (non multi-band), EQ and Limiter, until you can hear and distinguish the results and need for more Mastering FX in your mastering chain.
Once you start slapping general modules on without understanding or hearing the need for whatever you are adding, the point is going to be lost in the non-understanding of why you should use certain mastering plugins.
The good and or great news is that you are in the digital domain, and experimentation is at a level that it wasn't really afforded in the analog days. So try, try and try again ...and get your listening and understanding of  what you are hearing and or not hearing skills up (takes time) and just remember to save before experimenting. You'll get there in time with persistence and sharpening your, what you are hearing and how to get what you want to hear skills up. Just takes time, effort and research, like you seem like you are doing.

It's a simple four-step program...
 
1. Find out what compressors your favorite record producer is being paid to endorse, and buy them.
2. Find out which compressors faithfully emulate the hardware your favorite producers of years gone by used, and buy them.
3. Post on KVR "what's the best compressor?", write down the 150 recommendations you'll get and buy them all.
4. Use every preset and struggle quietly as you try to figure out which one will reveal the promised pixie dust.
 
5. (Optional) Study what compressors actually do so you know what to listen for, then go back to KVR and sell all your compressors to the next batch of frustrated seekers.
 
Seriously, though, compression has basically two uses: envelope shaping and leveling. Every application is some variation on those two things. Start with the simplest, bus glue. It's a form of leveling, wherein the loudest moments are quieted and the quiet moments are made louder. And despite internet lore to the contrary, you can do it with nearly any compressor.
 
The main variables are attack and release times, which determine how aggressive the effect is going to be. Do lots of A/B comparisons as you experiment with these parameters. Use large ratios (8:1 or higher) to start with, so it's easier to hear what's happening, and then gradually back the ratios down to gentler values (2:1 or less).
 
When dipping your toe into the multi-band world, start with just two bands crossed over at ~200 Hz and note how it differs from a broadband compressor, not just in the low frequencies but also in the high end. Listen to high-frequency percussive components such as tambourines, shakers and ride bells. Note that they remain more distinct as you turn the volume up when lows are compressed independently of highs.

Maybe get SampleMagic's Magic AB or the free Nugen AB Assist (http://connect.avid.com/NUGEN-Audio-plug-in-download.html) to compare with tracks you like.

I agree with the approach Mike Stavrou does. Here is basically what it is. I have posted this before. I find it really works.
 
How to set up a Compressor (properly)
 
A good tip to get a robust good quality sound out of a compressor is setting the parameters of a compressor in the right order otherwise you will never get it right. You need it to do a certain job and want the signal to still maintain a higher integrity. Putting it the other way a compressor can also make your signal sound bad.
I got improved results by doing it this way. If you do this you will get much better sounding compression on your tracks or mix busses. Think ARRT. Attack first, then release then ratio then threshold last. Here is how you do it. This works equally well on individual sources as well as busses and final mixes.

Lets set Attack first.

Start by setting the ratio very high, release as fast as it will go. Now feed audio in and lower the threshold until the compressor starts kicking in and you can hear it. Listen to the attack part of the music. The leading edge or attack transient. While the Attack setting is fast the compressor will jump all over the music and literally destroy the front edge of the sound. Slow the attack down slowly so the desired attack transient is achieved. Listen to how even as you start to apply small amounts of attack the music or the transients all come back and start to sound good again.

(Note: People often have the attack setting too fast and the compressor is destroying the attack part of the music. No matter what you do with other parameters the music will always be screwed) Setting the attack is very important for final mixes as well. Too fast and once again the music is destroyed. The ultra fast release lets you hear far more individual attacks than a slower setting.

Now we set Release

Release controls the speed at which the sound glides back after you have punched it away. The idea is to get that speed to become a musical component of the sound. Think how slow can I get it while maintaining some control. The power in a groove is in a slower moving wave. Listen to Release and feel the way it bounces back at you and at some point it will be like a swing.  This time does not necessarily have to relate to quarter notes or the groove in the music so much but an overall release groove.


Next we set Ratio

The idea is to lower the threshold as much as you can without losing the effects you have created with Attack and Release. The higher the ratio, the smaller the sound is, although it is more controlled. The lower the ratio as in 2:1 it feels like a larger image. The idea is to find a ratio that sounds big but is controlled. Listen to the size and firmness of the sound.

Last we set Threshold

It is desirable to set the threshold so that the compressor is not compressing all the time (Unless that is the effect you want of course) The correct setting will see the dynamic movement coming to rest at special moments. Too low a threshold creates a flat lifeless sound. Permitting the dynamic movement of the sound to come to rest in some quieter moments allows that moment to attain a momentary bigger 1:1 presence and prevents it from rushing towards the listener with unwanted noise. It is bad enough that quiet moments are small without being squashed smaller still due to higher compression ratios. Each time a sound comes up for air it attains a sense of reality a 1:1 ratio.

Makeup gain or output level

Compressors by their very nature are attenuating the signal so the output needs to be put back to where the input signal level was originally. Use the gain reduction indicator to give you some idea. ie if a compressor is providing mainly - 3db of gain reduction for half the time then set the Makeup gain to +3db to get the signal level back up to where it was.

Compressors are not black art or require rocket science. They are simply one of those processors you need to be careful with and have an approach to setting. Using this method above will give you great transparent sounding results. Even cheaper compressors will sound better after doing it this way. And even if you adjust them differently to this method, try it and see how it sounds after. You might be surprised.  One of the problems is that you cannot setup a compressor fast and move on like you can with other effects. The compressor takes time to get it sounding right. The moment you insert one you have to be prepared to put the time in to get it right. Don't rush it!
 
 
 

Jeff Evans
...How to set up a Compressor (properly)...

 
Jeff,
 
Thanks Jeff! That's the best method I've read about setting up a compressor and have pasted it into my ONE NOTE application as a reference!
 
Walt
 

Did two.