Are you operating in the correct SPL window for high-fidelity listening?

I have found the best volume is highly system dependent and that the  volume is higher on inexpensive (less refined) systems. The better my system, the lower the volume I listen to music in general. Now 65db to 75db. I think I started in the high 90’s back in the 1980’s. I don’t thing this is just and age thing, well after 30 or so... but there could be a bit of that. 

I got interested in this "the correct volume" idea. I had season ticket to the symphony for a decade (7th row center). So I would carefully listen to orchestral pieces containing sounds emerging from the silence and crescendos that would confuse my hearing (this is when your ear drums are overwhelmed by the volume). Then I would go home and set my system for that volume. I’m not sure how you could say anything meaningful about average volume of a symphony. I will listen to symphonies at that volume so creshendos are similar (a bit lower than live) to the orchestra hall. But most music I listen to at lower volumes. I seldom find rock or jazz more satisfying at higher volumes unless it has some tie to when I was a teen/early twenties and craved really loud music. 

Not just how loud it is but the position of the volume knob in order to achieve the level of loudness is very important. It's called attenuation.

Another way of looking at this... when you turn down the volume knob you are appling the brakes to REDUCE the signal (resolution). When you turn UP the volume you are releasing the brakes.  The only job of the volume knob is to attenuate/restrict the signal. The less interference the better.

When you have a preamp that adds too much gain and/or place it with an amp that more power than you really need it creates a situation where the volume knob is very low and restricts the signal coming in. It takes an attenuated reduced resolution and then boosts using gain to make it louder. This gain combined with lower resolution adds distortion. This can lead to listener fatigue.

A properly matched system allows for a higher volume knob setting. This allows for LESS attenuation and better resolution. 

A volume knob set at 8:30 is much too low. The higher, less attenuated it is the better, provided it gives you the proper DB of loudness without clipping. 11 or 12 is much improved.

First, the problem is how you measure SPLs.  The measurement will vary greatly according to how you aim the sensing microphone and what type of speaker you are measuring.  For example, planar (either electrostatic or electromagnetic) speakers are characteristically different from multi-driver electromagnetic cone speakers in how they do or don’t focus sound. Also, the microphone sensor of your SPL meter may have a very different angle of admittance than mine, etc. Second, what gdaddy wrote would seem to apply to digital components with a volume control that operates in the digital domain.  Such volume controls do lose resolution, the more they attenuate. I think it’s because digital volume controls simultaneously reduce the bitrate. Most modern digital gear has been engineered with excess bits, so controlling volume in the digital domain does less or no damage than previous. I am not aware, nor have I ever read that the same applies to an analog volume control. Yes, they attenuate, and yes our hearing acuity for very high and very low frequencies will vary with SPLs according to the Fletcher-Munson curve, but why would you lose "resolution" (except for the extreme frequencies where it is more a matter of detecting their presence with your ears) when attenuating? Also, analog controls may be of many different types. Most common are series, ladder, and shunt types.  In a series type control, there are resistors in series that add up to produce the desired level of attenuation.  There you may lose some resolution because of noise and inductance inherent to passing the signal through a string of resistors soldered together in series.  In a ladder type attenuator, for each level of attenuation there is only one resistor in series and one in parallel with the signal.  With a really good ladder attenuator, there ought not to be a problem, Shunt attenuators operate most like ladder types but have some potential issues related to input and output impedance.

85-90 db is a good target.   The most lifelike and still safe for ears.  
 

I shoot for that generally for serious listening and use a db meter to help. 

Some recordings are "cut" hotter than others. There is also the ambient noise of the listening room to take account of, leaving to one side the noise threshold of the system itself. Using high sensitivity speakers (104db), I had to work like the devil to quiet the system from inter-component grounding anomalies and to ensure that the power feeding my system was quiet without using "conditioners." All that said, I can hear the program material at low volume, but to energize the fairly large room my main system is in, I do need to give the system some gain.

I've mentioned this before, but every recording seems to have a "natural" volume at which it sounds best. I manage to avoid having the system "play at me" or sound like a reproduction system; instead, I'm able to get a pretty organic sound that appears from a silent background. Granted, there is considerable variability in different records; but I don't let sonics dictate my preference of listening material. Most of what I'm listening to is small combo jazz on records manufactured in the early-mid '70s, which was a nadir of vinyl production in the U.S.