The sensitivity is an average number, and it can be off by quite a bit.  Many things can contribute that aren't taken into account by the simple number you get when looking at speaker specs.  Things like room reflections, enclosure gain (ported box or horn load can change sensitivity at the lower frequencies by 3 to 6 db and if you ever hit a resonance point, all bets are off).  Most speakers will offer a graph of output volume at a fixed distance verses frequency when the speaker is given a fixed amount of power.  You should be able to see a correlation between that graph and the sensitivity number.  But even that can be off as many times the speaker is driven by a fixed voltage (2.828 V) and to get power you need to use the speaker impedance which also varies by frequency, even though the number you usually see is fixed (8 or 4 ohm).  That is also an average.

Then there is the human ear to consider.  It is less efficient at lower frequencies that at mid-range frequencies.  Pink noise takes this into account by putting more energy into the bass to make it seem like all frequencies are equal volume.  White noise is all frequencies at the same volume and that always seems a bit lacking in bass to the ear.

So, all these average numbers can really only be used as a ballpark starting point or in relative comparisons as a 94 db 4 ohm speaker really ought to be louder than a 91 db 4 ohm speaker - but even there you need to be careful (is it measured at 1w/1m or 2.83v/1m)

Dan

power = voltage squared divided by impedance.  So if the speakers are 8 ohms, 2.83v is 1w.  If the speakers are 4 ohms, 2.83v is 2w.  Obviously a 91db 4 ohm speaker driven by 2w should be louder than a 91 db 8 ohm speaker driven by 1w.

I really have no idea how the speaker sensitivity number is derived.  But I would hope that a number of different frequencies are measured and some sort of average is taken.  If I had to guess, being the pessimistic person that I am, I would imagine that the speaker manufacture comes up with a way to measure his speaker that makes it look the best.  A few years back a "new" series of raw drivers from Madisound appeared in the marketplace that had sensitivity numbers about 6 db higher than anything else comparable.  The one that sticks in my mind was a 15" woofer / subwoofer, 8 ohms, rated at 103 db 1w/1m.  Once purchased and installed the they sounded good, but the actual sensitivity was closer to 96.  A few years later these drivers from Madisound disappeared from the marketplace.  You can find them every once in a blue moon on Ebay.

Dan

Is the sensitivity of the measured frequency ranges(300hz - 600hz) a good representation of the other frequency sensitivity?  I guess, maybe? Otherwise they would measure across a wider frequency range?

To add more to the above - speakers are dynamic analog devices. Just look at the impedance curve at different frequencies! The goal of a good design would be consistency, but being a dynamic analog device...well, you cannot change the laws of physics, captain!