Probability

If the selection process is truly random the probability in each "draw" isn't affected by anything that happened in the past - it's like a coin toss, chances of either head or tails are 50-50 no matter what happened in the previous toss.

But your question really is - staying with the coin example for simplicity - at what point in a repeating sequence do you start wondering if the coin is fair? It is theoretically possible to get 10, 20, however many times the same result (head or tails) in a row, so traditional probability calculus doesn't help. There's a new branch of mathematics dealing with phenomena asymptotically approaching zero or infinity, makes for pretty graphics >

http://www.technologyreview.com/blog/arxiv/27656/?p1=blogs
> but realistically if I'm betting and I see a coin come up anything to make me lose more than 3 or 4 times in a row I'll get hold of the coin and examine it. Your chance of getting picked at random is 2/70, if it happens again this year, worry! Or, as James Bond always said, once is happenstance, twice is coincidence, third time is enemy action.