Well, defining "beneficial" would be a good start.
I can think of two reasons why, in many cases, an array would be a preferred storage solution. One being logical organization --- an array is superior to $n related variables in the same way that a file folder full of job applications (particularly if sorted) is superior to a group of them in piles on my desk (not that I have many job applications, nor employees....)
Secondly, and what I've seen most often in various writings, is simply effort and therefore time in data access. If you can use a more general function to grab several data instead of just one with each db access, that's inherently superior in most cases. I suppose that, in reality, each programming action should be subjected to testing for its impact on such things. One could at least conceive of a situation in which a Real Fast (TM) database could actually make a script's "bottleneck" be the server's memory capacity (e.g., amount of memory available to play with your arrays). But most people assume that things will work the other way 'round....
