Err...if we assume that children perform according to usual Gaussian distribution curve and skewness of probability distribution is zero, 50% of kids WILL be below average...it's a simple statistical fact.
Otherwise, how could they calculate "the average"??
