I'm not sure what you mean by "the way the game works in general." All dice rolls are simply a question of shifting the distribution curve of results, and I assume GW's designers are sophisticated enough to treat them that way. In fact, I see more evidence of the designers branching out mathematically these days; consider the various curves of the Punisher as opposed to the Exterminator.
If by "makes sense" you mean can be justified from a fluff standpoint, I personally prefer the multiple test position for the following reasons:
- It eliminates the weirdness of multiple units being more effective than single units by virtue of being multiple units. Consider, for example, two squads of five snipers lined up end to end. Under the single-test model those two squads are more likely to pin an enemy unit than the same ten models lined up end to end in a single unit. Similarly, three Basilisks in a single squadron are somehow less effective at pinning an enemy than the exact same three Basilisks, in the exact same positions, organized as three separate units. This doesn't "make sense" to me from a fluff/real-world standpoint, and the multiple-test model avoids this weirdness.
- Shooting only happens simultaneously from a rules standpoint. We don't know how long a turn of gameplay or even a single shooting attack takes, but it's intuitively obvious that shooting attacks take time. As an easy example, if a single bolter kills two models, it's obvious from the fluff nature of the bolter that those two models were not hit at the same instant. Similarly, a squad of ten models is unlikely to fire at exactly the same instant. A series of multiple shots ringing out, or a series of shells falling from the sky, over a period of time seems to me like it would create more uncertainty among the victims than a single instant.