It would be good to see how subjective impressions match up with objective measurements, but unless the listening tests are blind, they won't have much scientific merit, due to expectation bias etc. You could eliminate one source of expectation bias at least by doing your listening tests before seeing any measurements of the particular speakers you're testing, so your subjective impressions aren't influenced by objective measurements (the same goes for your DAC/amp etc listening tests).
Thanks so much for expanding your tests to speakers, it's a monumental project that will be of huge benefit to us consumers.
I agree that listening before measuring would probably be best, but I don't think Amir should be aiming for that because it can't really be done with one person and it's a huge time sink.
Moreover, there's expectation bias any way you cut it. From my experience measuring speakers, either you risk making your listening impressions match objective data, as you mentioned, or you can fall into the trap of doing the opposite and try to make your analysis of the objective data fit your listening impressions.
Even with full anechoic data, there's still room for interpreting the graphs and focusing on different details. Of course, the great thing about amir making this data public is that we can interpret the data for ourselves too
Just FYI, the types of speaker measurements Amir is proposing to make have a very good correlation with listening tests.
While true, the correlations are with blind listening tests of multiple subjects, which I think is what bobboo was getting at. But yes, overall impressions should be in agreement. While I've had a few moments where on and off-axis data didn't seem to line up with my impressions, the vast majority of the time everything I hear can be explained by frequency response graphs or SPL limits.