u can see which heatsinks are louder in the enclosure in relative terms to the other one's that they test
That's the point. YOu can't really see if the differences get minimized by boundary effects. (Also, I'd be surprised if they get much more than ~10 dBA attentuation in the box. You'd be surprised what it takes to block out noise!)
If all the under 500Hz readings are boosted by say 10 dB, it has the effect of compressing the differences between the HSF. Sort of like "FM radio announcer voice" effect -- some mics (studios, stations, whatever) deliberately add a huskiness to all male voices (by boosting certain bands in the lower midrange and low frequencies). As a result, they all sound kind of similar - rich warm sexy voice.
That's the kind of effect recording in a small box has.
I still think it's great that they try, and it does give some idea of noise levels.