Complicated issue
The answer to this isn't easy nor terribly intuitive. First, let me state that, from a risk-averse point of view, this is the quintessential play, which accounts for the bulk of all the edge that can be acquired from using a slew of R-A indices. (Note to laotse: one index, two indices. There's no such word as "indice.")
This is a play that we say is not very volatile as the count changes. By that, we mean that the expectations remain rather stable, with a rising count, such that the change in e.v., from one TC to another is very, very slight. At TC = 0, the player edge from hitting (as many times as he wants) is a very slight 0.03%, while the edge from doubling (one card only) is ever-so-slightly negative, at about -0.01%.
As the count rises, the edges for BOTH hitting and doubling each rise, but because we're betting twice as much, when we double, the double-down e.v. eventually catches the hitting e.v. at +4 and then passes it, as the TC goes even higher still (we need +7 to make the R-A play).
Consider, also, that the player has a greater chance to make 21 than the dealer, because the dealer has already checked that he doesn't have a natural, whereas the player has, roughly, one chance in 13 (7.7%) to catch an ace on the double. The dealer, on the other hand, turns his ten into 21, with multiple hits, less than half that percentage.
Finally, although it is true that IF the dealer turns out to be stiff, he will break more frequently with the higer count, unfortunately, he actually breaks, overall, LESS with a ten up at high counts, because, of course, his chances to be stiff in the first place decrease with the rising count.
All told, this is not a very advantageous play to make at any count, and if you never doubled 10 v. 10 for the rest of your life, you wouldn't be missing very much at all.
Don