IMHO loops appear quieter because noise and signals are a little lower than a dipole.
The S meter reading of the noise on an antenna is pretty much meaningless unless you also know the S meter reading of the signals. Signal to noise ratio is the only meaningful parameter when talking about noise. Antennas at differing heights have different patterns. When you have different patterns the signal to noise usually changes. That goes for dipoles as well as loops. Noise to an antenna is the same as a signal. It has no way to discriminate between the two. The only affect will be due to differing patterns.
So if you are trying to compare signal to noise on a dipole versus a loop (assuming the loop is oriented vertically and horizontally polarized), the height of both antennas need to be such that they both have the same gain, and both need to be oriented in the same direction, and far enough away so that they don't interact. Then their patterns will be almost identical, which of course will produce the same signal to noise ratio (or so close you probably can't measure the difference). A locally generated noise source close to one of the antennas can skew the results.
The top wire of a loop needs to be higher than that of a dipole to have the same gain. If you make the top wire the same height for both, the loop will probably have lower S meter reading on noise due to lower gain, which is probably why most people say the loop is lower noise. Signals are also lower under that condition.
There is one special case where a loop is quieter, in the presence of precipitation static.