Could someone explain why I'm seeing around 1.24% rejected proofs of work on the live stats page?  What causes this and what can I do to reduce this number?

Thanks,
Bill