A new process is being considered for the liquefaction of coal. The old process yielded a mean of 15 kg of distillate fuel per kilogram of hydrogen consumed in the process. Let μ represent the mean of the new process. A test of H₀ :μ ≤ 15 versus H₁ :μ > 15 will be performed. The new process will be put into production if H₀ is rejected. Putting the new process into production is very expensive. Therefore it would be a costly error to put the new process into production if in fact it is no better than the old process. Which procedure provides a smaller probability for this error, to test at the 5% level or to test at the 1% level?