In some positions you have a choice, there will be multiple ways to pursue an advantage. In others of course there is no choice and you must attack or you must go for the superior endgame or something else.
Also winning a game of chess is more than finding the best move in a position. As you noticed you have to follow that move up with good moves. So other than style sometimes a player chooses the method that is easier. E.g. maybe an attack wins, but you have to find a series of "only" moves but the endgame wins with any combination of reasonable moves. So of course as a practical matter you'll choose the easy way to win and not the 8 move mate the computer happens to see.
And although that's an easy example, there are many cases where it's more subtle but the computer hasn't chosen a practical way to play. e.g. trust the computer to spot your tactical mistakes, don't attempt to interpret the ideas you should have chosen unless you're near master strength... often because of the search depth the computer won't even register these other ideas unless you work with it a bit (play out good lines to help it see, give it lots of time to think, analyse with multiple engines, etc.) and that takes a lot of general chess knowledge.
All that said, my advice to you if you feel like you're leaving attacking advantages unused is to get a game collection of Alekhine, Tal, Nezhmetdinov, or Shirov or a book like Art of Attack and learn what elements in a position are necessary for an attack to work as well as how to conduct a successful attack. The computer will never tell you this information.
To answer your topic title though... to quickly improve do tactics. If you're all tactic-drilled out then there is no more quick fix :) To get better takes work. I highly recommend Pachman's Modern Chess Strategy to give you the basis for good middlegame play.
And by the way, going into a winning endgame isn't "blowing your advantage." A winning position is a winning position.
I feel like I've hit another huge roadblock in chess; I play against deep fritz often on my iPad on 1500 and usually I lose. That's not so frustrating, after all I usually learn something even if meager every match.
The frustration comes from analysis when I see my attack was actually 3 to 6 pawns advantageous (according the fritz) and I squander it with a series of innacurate moves. Beating the computer in an endgame is very difficult and at my level I feel that it's okay if I mess up there ( and often I do) but it's the fact that I can't see combos through that really gets to me.
It even happens when I play weaker players that are friends and I end up settling for a winning endgame instead of throwing everything and the kitchen sink at them, which I usually am justified in doing in the positions that arise.
I think that humans can't find resources to defend as well as computers but I'm going join the Marshall chess club soon and I don't want to blow my advantage every time I get one.
My last thread about 17 months ago received very good posts and since then my game has improved a lot (for instance instead of playing the moves I learned I'll often deviate seeing my opponent neglecting my trumps from the opening) and while this thread will probably be less philosophical I think there are enough good players here to help me out. Perhaps I'm going about it the wrong way and I shouldn't attack unless I see the absolute best combination but again I feel that sort of chess isn't only boring but unsuccessful. For instance, if I move my queen towards a casted king dropping my D pawn it might lead to a collapse in my position if I don't play perfect vs fritz but that queen move is in fact the best move according to fritz if followed properly.
I hope I'm not being too ambiguous and that I can get some good advice again. Again if anybody wants to help but is still unclear what I'm asking I can clarify.
Thanks