To understand why a single vote per race tends to favor less centric candidates, consider a simple lab experiment where students in a class vote for their favorite marble. If five marbles are assigned names and are placed "up for election", and if three of them are green, one is red, and one is blue, then a green marble will rarely win the election. The reason is that the three green marbles will split the votes of those who prefer green. In fact, in this analogy, the only way that a green marble is likely to win is if more than sixty percent of the voters prefer green. If the same percentage of people prefer green as those who prefer red and blue, that is to say if 33 percent of the voters prefer green, 33 percent prefer blue, and 33 percent prefer red, then each green marble will only get eleven percent of the vote, while the red and blue marbles will each get 33 percent, putting the green marbles at a serious disadvantage. If the experiment is repeated with other colors, the color that is in the majority will still rarely win. In other words, from a purely mathematical perspective, a single-vote system tends to favor a winner that is different from the majority. If the experiment is repeated using approval voting, where voters are encouraged to vote for as many candidates as they approve of, then the winner is much more likely to be any one of the five marbles, because people who prefer green will be able to vote for every one of the green marbles.