I used Tapio Lokki's 2019 and 2016 paper on listening room, concert hall acoustics, and listener preferences to idly speculate on a model of listener preference and loudspeaker characteristics: proximity/clarity (high directivity conventional loudspeaker and/or absorption of early reflections), width/envelopment (wide directivity conventional loudspeakers and/or preservation of lateral reflections), reverberance/spaciousness (omnidirectional, possibly bipolar speakers splitting the difference between this and clarity/proximity), timbre (many critical listeners of classical music seem to like wider baffle speakers, and my speculation is that relatively constant, rather than smoothly rising, directivity between several hundred and several thousand hertz may contribute to this), and bass. These characteristics aren't necessarily mutually exclusive--for example, Earl Geddes called for extremely high directivity speakers to be used but with strong later contralateral reflections--but listeners might weight the individual characteristics in terms of relative importance in consider loudspeakers and listening room setups.This is definitely true for me. Rating purely the "Spacial" component, I rate wider dispersion higher in mono, and medium or narrow dispersion higher in stereo.
For mono listening, I prefer the spatial presentation of my Revels far more than I do my JTRs(narrow dispersion) or Genelecs(medium dispersion). The Revel throws a wider soundstage and disappears much better, which are the 2 components of how I rate "spatial" quality from 1-10.
In stereo, I actually rate the Revel the worst of the 3. The Revel still disappears a bit better, and it still throws a wider soundstage, but, it now does worse on the (new) third component, which is the tightness of the phantom center image. This may just be a personal preference, but I love tighter phantom imaging more than I love wider soundstage, excluding some types of symphonic music. For certain types of symphonic music, I actually enjoy a more diffuse phantom image, as it sounds closer to what I hear at live symphonic events.
In your case, I correlate proximity/clarity with preference for "tight" or "pinpoint" imaging. I even wonder whether this preference may correlate with narrow baffle or smoothly rising directivity, but I have no evidence to support that.
I might idly speculate that the wider directivity enhances the perception of width/envelopment, resulting in less pinpoint localization of "the speaker is exactly right there."A year or two ago I blinded the Genelec 8030c against the Revel M105, but only in stereo. I really wish I had done a subsequent mono test to see if the results changed. The Genelec won the stereo test, but I'm pretty confident the Revel would win a mono test. The reason I think this is twofold:
1. Tonality is almost identical, so it's gonna come down to the "Spatial" rating
2. The Genelec slightly won(imo this is why it won) because of the spatial component, which was due to the tighter phantom image it throws. I'm somewhat guessing this is why it won based on why I prefer it, and comments from the listeners. A common comment in favor of the Genelec was something like "this one sounds more like the singer is there in front of you". In mono, though, I actually rate the Revel higher for its spatial representation. There is no phantom image anymore, and the narrower dispersion makes it clearer that the sound is coming from a speaker right in front of you, which means the Genelec doesn't "disappear" as well as the Revel.
The Beolab 90 would first have to be shown to be virtually identical between directivity modes. I personally believe that the best current way to begin to address the questions in the way that you're outlining would be with simulated listening environments (and simulated loudspeaker directivity indices of varying curves) using a setup like the Aalborg array outlined here.One thing that may be the cause of some difference is: "What question are you trying to answer with this blind test?". Are you trying to find which speaker sounds best with "average" placement/toe-in? Or, are you trying to find which speaker sounds best with "optimal" placement/toe-in? I think the Harman research was aimed more at answering the former, whereas my tests were aimed more at answering the latter.
For answering the first question, the way Harman does it(speakers in same spot with same toe-in) makes the most sense, as user room placement and toe in will likely be all over the place. Doing it like this does definitely bias the results towards wider dispersion, and Dr. Toole has even mentioned this too. But, Dr. Toole also brings up the great point that this bias is not really a bad thing, as placement insensitivity is an inherent advantage of wide dispersion. My Revels sound great without much setup at allb at almost any position. Sure, you can optimize it a bit(maybe 10%?) by messing with toe-in, but they pretty much always sound great. My JTRs on the other hand are much more finicky. They go from sounding bad to sounding amazing just by adjusting the toe-in a bit.
For answering the second question, you would need to first find the optimal position and toe-in for each speaker under test, and then design a machine that can quickly place each speaker in that optimal position and orientation. Such a machine would likely be much more expensive and complicated, and maybe isn't answering the question that is most important to a manufacturer. The way we handled this was by spending a few days prior to the event finding the optimal placement/toe-in for each speaker, and then marking those positions with color coded tape for each speaker. On test day, when the listener called for the switch, we had two people to move the current speakers out of the way, and two people to move the new speakers into the color coded position/angle. It seemed to work really well, but requires a good number of people, and also our switch times weren't 3 seconds or less like the Harman switcher is(I'm hoping the ABX comparator can assist with this).
Anyways, glad to see Dr. Toole participating here, and this is something I'm super curious to see more research about. A loudspeaker like the Beolab90 would perhaps be an excellent test subject. Maybe the "wide" or "omni" mode is preferred in mono? Maybe the "narrow" mode is preferred in stereo? To truly answer this mono vs stereo debate for myself, I need the ability to hold the FR component as a constant, and let dispersion width be the only variable.