The future for hybrid and digital

Hence if we want to fix &dquo;the&dquo; probability of an aerror to the value a, then we compare each of the two F-statistics (El E2) with the F-value tabulated for a/2 (and obviousiy vi and v2 as in Equation 10). We reject the null-hypothesis if F, and/or F2 are significant. This approach is again conservative since Equation 13 shows a -< sign, not an = sign as in Equations 11 and 12. (Moreover, the power of this approach is decreased since the degrees of freedom are reduced.)