Abstract:

ESDU 93023 uses second segment climb measurements to provide simple examples of statistical significance testing and confidence levels. Certification trials of a twin turbo-prop aircraft established that second segment climb gradient exceeded scheduled by a mean margin of 0.09 per cent with a standard deviation of 0.35 per cent. The comprehensive nature of the tests allowed the values to be taken as applying to the population. Two sets of ten points were then obtained, each set with a variant of the aircraft differing from the basic design by small aerodynamic changes. Use of Fisher's Variance Ratio F test established that the standard deviations (variances) of both sets did not differ significantly from that of the population. The means of both samples were then compared with that of the population using the Normal cumulative distribution function to establish the confidence that the differences arose by chance. In one case the difference is seen to be significant and the performance had to be rescheduled. The necessary changes are calculated, and the probability that any single test point might fall below the new scheduled level is established.