50/50 is statistically the most likely outcome when tossing a coin. If you flipped a perfect coin (in a truly random way) 1,000 times, this does not mean you will get 500 heads and 500 tails. In fact the more times you flip, the less likely an exact 50/50 split becomes i.e. if you flip 1,000,000 times, the chances of 500,000 heads and 500,000 tails are a very unlikely result. However, the more times you flip, the closer you will get to 50% heads and 50% tails. For 1,000,000 flips, the chances of being within the range 499,500/500,500 are just over 68% and the chances of being within the range 498,500/501,500 are better than 99.7%.

Statistics, like these, are often used in science to check the validity of experimental results which, by their nature, involve measurement and some uncertainty. The statistics are used to verify that the experimental results are consistant with a theoretical prediction. It is in the nature of science that everything we know as "laws" are based on theory that can only ever be consistant with experiment. If any experiment shows that the theory's predictions are outside the statistical range that can be accounted for by experimental errors, then the theory is wrong and the science has to be re-thought through. An example is in showing Newton to be wrong and Einstein to be right in prediction of the decay of mu-mesons. The fact that Newton was wrong is not really to do with statistics but due to not having a full understanding of the underlying physics. While Einstein may also be wrong, this is not because of any inability to measure with perfect precision. The statistics are an aid in knowing the precision of measurements and then adjudging the validity of the data used in experiments, and not at all something that invalidates the science itself.

Quantum mechanics blurs this distinction somewhat because the statistics involved in measurements are part of the physics itself. But this does not make the physics any less valid or correct; it is just that we have to deal with probabilities of a predicted outcome more than within classical physics.