The "A" weighting is key. "A" weighting greatly discounts low frequencies, and capsule noise is 1/f, so it's predominantly low-frequency. It's generally so low that you can't hear it except in special situations, so in a general sense the weighting is doing the right thing.
But the dB specifications that result from its use shouldn't be compared too closely. The "A" weighting curve is a simplified version of the ear's response at 40 phons (a measure of loudness which equals 40 dB SPL at 1 kHz). At much lower levels such as this, the weighting would need to be rather different in order to reflect what we actually hear, so it's not an especially appropriate way to specify noise in studio microphones. However, it's well-understood, and test equipment is readily available that implements this weighting curve, so the industry stays with "the devil it knows."
Equally important is the fact that noise isn't all steady and smooth. To the extent that there are spikes (impulses) in any noise, its subjective level will increase even though the long-term average power may be virtually unaffected. And so-called "RMS" values for noise are nearly always used whenever you see "A" weightings. This implicitly represents a longer-term average which tends to smooth away impulses--so that's a further layer of "lying by means of statistics."
The CCITT (or sometimes CCIR) quasi-peak method takes impulse noise into account and uses a weighting curve that is more realistic as an approximation of what we actually hear (or don't hear) at such low levels. Many manufacturers don't publish that specification, however, since it's typically 10 - 12 dB higher than the "A"-weighted RMS noise value. Giving an "A"-weighted RMS value as the only noise specification is like publishing a polar diagram at 1 kHz without showing what happens to the pattern at other frequencies.
Different manufacturers mean different things by their specifications in general. To some extent this reflects cultural and historical differences. To some manufacturers, particularly in Germany, specifications are very explicit, with standardized methods of measurement, and they are essentially part of the guarantee. Those manufacturers have a long history of building microphones for professional broadcasters and recording companies where the engineers are university graduates with engineering degrees--people who test and measure all the equipment that they buy. Those companies generally will repair a microphone at their own expense if it doesn't meet their published specifications within the guarantee period, and sometimes even after it.
To other manufacturers, particularly in China, specifications represent an aspirational category: They are asking you to think of their product along with microphones that really do offer such performance, even though their microphones don't, and they know it full well. In some cases, when they copy the look of a European microphone they also copy its specifications verbatim, or the specifications of some other microphone that they would like to compete with--e.g. one big Chinese manufacturer's Web site displays the polar diagrams of the Schoeps MK 4 V for a completely unrelated and dissimilar microphone of their own. That can't have happened through accident or oversight.
The American approach seems to be about midway between those two extremes--U.S. manufacturers are always a little surprised when someone takes their specifications seriously, but if you do take them seriously, they usually will, too, which is nice when it happens.
--best regards