02-19-15: Effischer+1. As I've said in more than a few threads here in the past, the main usefulness of specs and measurements is in identifying and **ruling out** candidates that would be poor matches with other components in the system (e.g., impedance incompatibilities, mismatches of gains, sensitivities, power capability, etc.), or with the listener's requirements (e.g., peak volume capability, physical characteristics, perhaps deep bass extension, etc). But not in selecting among candidates which make that cut, where listening is essential.
Being rather partial to my money and not enamored of do-overs, I insist on an audition before buying. That said, much can be inferred from specifications, and they are the tools I use to determine what to audition.
By doing that preliminary screening based on specs and measurements, the randomness of the selection process is decreased considerably, together with the likelihood of expensive mistakes.
A second major usefulness of specs and measurements is in diagnosing problems or sonic issues that may arise or become apparent in a system that has already been assembled.
It is also sometimes possible, btw, to rule out candidates from consideration on the basis that some of their specs are simply TOO GOOD. A classic example being THD numbers that are near-infinitesimal fractions of 1%, which is often indicative of heavy-handed application of feedback, resulting in TIM distortion and excessive amounts of the most objectionable kinds of harmonic distortion. Or specs that are so good that they may be indicative of misplaced design priorities, and compromises in other aspects of the design.
As evidence of these usefulnesses of specs and measurements, I couldn't begin to count the number of threads here in which I and many others have found it useful and necessary to refer to the measurements John Atkinson provides in Stereophile.
Regards,
-- Al