A random survey of 205 drivers measured each driver's "anger" score. Higher scores indicated higher propensities for road rage. The average anger scores were calculated for drivers aged 16-25, 26-35, 36-45, 46-55, and 56-65. A one-way analysis of variance was performed to assess whether or not there were differences between anger scores of drivers in the different age categories. Using the F-distribution table, what is the appropriate decision rule that we should use for the global F-test if we are interested in maintaining a type I error rate that is less than 0.05?