There are quite a few sites for training in chess tactics. For those having experience with using any of these, how did these affect your rating? Is there a strong correlation between e.g. http://chess.emrald.net rating and ELO? Has the use of these sites made a significant change to your ELO?
As BlueRaja said, chesstempo is good, but you can train tactics @ FICS, chess.com, gameknot.com, chesscube.com...etc
Famous quote "1 problem a day keeps the patzer away" Indeed when you start doing tactics regularly you immediately see results. You tend to do a lot less blunders (which obviously takes you a rank up), remember some positions and tactics.
There's also some tactics training that I will describe as an answer to your other question.
Tactics are the most important part of the game, so training in tactics will improve your game the fastest.
Other parts (opening, endgame etc.) are helpful, but they take more effort to learn well. Tactics gives you the biggest "bang for the buck."
I haven't seen a site that offers the right kind of tactics training for beginners, including emrald.net. While I think strong players do well on sites like emrald.net, I do not believe the converse is necessarily true: that players who only do well on sites like emrald.net do well in playing real games. Strong players certainly have excellent tactical vision, and high ratings on emrald.net will reflect this. But strong players also possess other qualities which make them strong, including a consistent and thorough thought process, which emrald.net does not measure.
For beginners and improving players, it's very important to start tactics study with a set (on the order of several hundred) of fairly simple tactical positions, and to go through them repeatedly, over and over, until you can instantly recognize them, spending only up to a minute or so on each position, then looking at the answer (and decreasing this time each pass through the set), and moving on to the next problem. Recognition is the goal in the beginning, not building tactical analysis skills.
I think the best way to do this is to use randomized "flash cards" with the only instruction being "white/black to move." One excellent source of suitable positions is John Bain's book: "Chess Tactics for Students" (as recommended by Dan Heisman, http://danheisman.com, the top guru of beginning/improving chess instruction). I copied, cut, and laminated all the positions in that book, making flash cards out of them, and studied them in randomized order repeatedly until I knew the entire set on sight instantly. This gave a huge boost to my playing performance and to my further tactics studies of more involved problems.
Tactical ability is certainly right at the top of importance, but along with it is developing a proper "thought process." One thing a proper thought process includes (among many others, this is just a simple example) is a safety check for all candidate moves (moves you're thinking about making when it's your turn to play): "will this move lose material or allow a mating attack against me?" Without a consistent thought process, all tactics, strategy, endgames, opening knowledge is basically useless. It takes just one bad move to lose a game!
There are many techniques to help reinforce a good thought process, but again, sites like emrald.net do not assist in that sort of training when used in the typical fashion. When I was using emrald.net actively, I worked only for accuracy, never for speed. Used in such a way I think the site can be useful as a training aid.