I mixed two songs for the same album.  One is a full country band.  The other is a ballad with just voice, finger picked guitar, cello—and very light bass and drums that come in midway through the song.
 
Both have master bass compression to keep the level maximized accodring to the master level meter.
 
The problem is that the ballad with light instrumentation sounds louder, possibly because the arrangement is so sparse that the relative level of the vocal is higher than on the full band song, even though both haved similar overall levels.
 
Now my question: do you normally mix a ballad with minimal instrumentation at an overall lower level so that the voice sounds consistent with the full band arrangements?  I could do that, but I don't want people reaching for the volume to turn up the ballad when it starts with just a finger picked guitar.
 
Thoughts on best practices?
 
 

the mix that sounds louder will have a higher rms value, if you can get mixes relatively to the same rms levels, then that should even it out more , i use this on my master buss http://www.meldaproductio.../mloudnessanalyzer.php

Thanks, Chregg, I will check it out.

Just using arbitrary made up numbers for conceptual purposes, suppose your overall mixes both have a level of 10.  Could it be possible that in the full band arrangement, the vocal volume level is 3 of 10 whereas in the ballad it is 5 of 10?  Maybe the user perceives the ballad as louder because the vocal is taking a larger share of the available headroom.

To put it another way, would it make sense to set the overal level by looking at the vocal level soloed and then build around that?
 
By the way, it looks like your Soundcloud link is broken, but I would like to hear your stuff. 

Konradh switch the metering to rms and peak, in both mixes, i dunno if you use busses for groups or anything , but from there you should be able to see wats 3 outta 10 or 5 outta 10 and balance it out more, if building you mix around the vox works, go for it !!! p.s ive scrapped my soundcloud account lol want something better

RMS levels will be vastly different in the 2 songs , for me it is best just to do this my ear. Getting the vocal levels matched
  is a good place to start. Put the song side by side and play the songs back and forth also listen how they transition. 
Adjust to taste, 
      
   

I'm not so sure. We are talking about music. I think one should respect the music. I don't see how a quiet ballad should sound as loud as a full on band performing with amplified guitars. True you want levels for a CD to not be too different from song to song but upping a soft song can destroy its mood. This can be handled with a careful fade out of the high level song going into the lower level song. But use a concert as a model where you start with a quartet and than go to a full orchestral performance. Should they both be as loud?   

We are in a world where the truth of the music is being manipulated and controlled by the recordest and masterer not the musicians. I think we need to let the musicians control how loud they want it.