What Etymotic (and you) are failing to account for is the difference in tactile bass percieved via speakers and headphones. The reason Harman's research found people preferred a bass boost when listening to music via headphones compared to speakers is because the former, even if they have very good bass extension, produce no (or at least very little) tactile bass, whereas the latter do produce full-body tactile bass (if they have adequate bass extension), which is closer to what you would feel at a live concert / club.
The bass boost people prefer in headphones acts as a perceptual substitute for this missing tactile bass, so subjectively speaking, the Harman headphone target is a perceptually more accurate simulation of what you would experience when listening to speakers, and so in turn gets you closer to a real live music experience. See
this study which showed the less tactile bass people experienced while listening to music, the more bass boost they preferred to EQ their headphones with.
I highly suspect that the reason Harman found people preferred an even higher bass boost for in-ear compared to over/on-ear headphones is because the latter can provide slight tactile bass cues on the outer ear / skull, whereas the former can't at all, so an even greater bass boost is required to compensate for even less tactile bass in the case of IEMs, in order to give a better illusion of a realistic full-body bass experience.