Depends on who you ask.
Manhattan showed me a study that basically proved each question is important - thus, take the same amount of time on each question. Don't just give yourself 3 minutes on the first 10 (thats really a lot of time, if you think about it, you'll be 5 questions slow by the time you finish the first 10, and thats assuming you time yourself correctly).
I suggest adapting your technique - if you see a question that you just know you probably wont get right - maybe you don't remember the way to solve it, take a stab at it, spend 1 minute on it, make an educated guess and move on. If you waste 3 minutes on it, you will probably just get it wrong anyway.
I saw one like this on the exam where, when I saw it, I instantly knew I was going to get it wrong. It was a ridiculously complex formula that would have required serious algebraic heavy lifting to figure out (it was a DS), I looked at the options, thought about how what I knew might effect the overall value, and moved on.