How would you Create a set of data using 72 trials that would prove two dice are "loaded", or not consistent with the expected values at the 10% level of significance.