My databases (mainly the one provided by SmartGo!) contain 17 games in which the pattern is played. 13 times white wins.
The statistics indicate that there might be a better reply then below. a was played 5 times and black won three of them.
This leaves white a winning percentage of 12 to 1 (!!) for a black reply at .
Not exactly. There are indeed 17 games where the approach is played, but (AFAIK) only 9 where the pattern is played (ie with black jumping to and the attachment plus jump upward by white).
I agree. However, it is not white's choice to play the side pattern. And according to statistics a better reply for black is to attach at a, in order to prevent the side pattern from being played.
No, it is indeed not white's choice. In fact the most common response by black after white finishes the Onadare is to make a low extension from his 4-4 stone along the top side, preventing a white approach from that side altogether.
The question of course is how reliable such statistics are for such a small data set. Statistically speaking again, professional players prefer to extend instead of attach by a margin of 13-5, apparently. I would personally think that the choice between these moves depends on the surroundings, and that it is impossible to say with certainty that one choice is better than the other based on such a small sample.