Our tests are actually closely built on the same algorithmic theory as the real test. You should expect to get many, many questions wrong - the vast majority of people, from 300 to 700, get half of the questions wrong. The test is not scored based on a traditional "percentage correct" model.

Your review of questions should be focused on ALL questions, not just the ones you got wrong. We often learn a lot more about the structure of questions and how to approach them from the ones we get right.

Especially at the 600+ level, you learn only marginally by just doing lots of problems. I would rather see you do one-quarter of the problems, but spend three times as much time going over them. On every question ask yourself:

What's the right way to approach this (and if it's a math or SC question, come up with at least two different ways to approach it)?
Which way do I feel most comfortable with?
Is there a more efficient way to do it?
Is there a more effective way to do it?
What's the wrong way to approach this?
(For verbal) Why are the wrong answers wrong? (Be specific.)
What are the traps?
How is the question trying to trick me?
How are the answer choices trying to trick me?
How would I make an educated guess on this problem?
What from the problem could I use on other similar problems in the future?
How will I recognize those other similar problems in the future? How will I know what to do?
How would I teach this to someone else? The "real" way? The educated guess way? How to avoid traps? Etc.

Until you can answer all of these questions about ANY problem - even one you get right in a minute, one that you totally understand - you haven't finished studying that problem.