Of all the variables that compromise speaker auditions, level mismatch is the most consequential and the least corrected. The louder of two speakers being compared will almost always be perceived as better — more detailed, more dynamic, more present — even when the difference is as small as 0.5 dB. This is not an opinion or a theory. It is one of the most thoroughly replicated findings in psychoacoustic research, and it applies universally regardless of the listener’s experience or awareness of the effect.
The reason is rooted in the auditory system’s sensitivity curve. The ear’s frequency response changes with listening level — low and high frequencies are perceived as quieter relative to the midrange at lower volumes, and more equal at higher volumes. This is the basis of the equal-loudness contours established by Fletcher and Munson. When one speaker is even fractionally louder than another, the louder one benefits from a perceptual shift toward fuller bass, more present treble, and greater apparent detail. It will sound demonstrably better to virtually every listener, regardless of its actual quality.
In a dealer audition, level matching is rarely performed precisely. The dealer switches between speakers using a selector switch, and even with volume adjustment, the levels are typically matched by ear rather than by measurement. A difference of 0.5 dB — well within the margin of error of casual level matching — is sufficient to systematically bias the comparison.
The practical solution is to insist on level matching before the comparison begins. Bring a sound level meter — a simple smartphone SPL app is adequate for this purpose — and ask the dealer to match the output of both speakers to within 0.5 dB at the listening position using a steady test tone or pink noise. This takes two minutes and eliminates the most significant source of bias in the comparison.
If the dealer is unable or unwilling to level match, be aware that the comparison is already compromised. A speaker that sounds noticeably better at a fractionally higher volume may sound identical or inferior at matched levels.