Randomizing speaker locations is needed to average out room effects.

Evidence

In any comparison test, if loudspeakers being compared are in different locations, additional tests are required with the locations randomized to average out the room effects. This necessity adds time and tedium to the test, so it is not always done. If listening tests are to be a much-repeated activity, this alone is motivation to create an apparatus to substitute different loudspeakers at the same location in the room, so that acoustical coupling to low-frequency room resonances and adjacent boundary interactions are constant factors. The improvements in test efficiency and effectiveness are profound—important to manufacturers and reviewers. — Toole, Floyd E, Sound reproduction- the acoustics and…, 89