B767Driver
New Member
Careful Jim,
You are using some "USA Today" info gathering and statistical analysis above. I could use the same analysis to prove that blue uniforms are safer than black ones...and that blondes are safer than brunettes. This is the type of analysis that gives creedence to the fact that liars figure and figures lie.
What is the statistical significance of the facts? Using a statistical test such as a chi squared or mann whitney test use you can determine the level of correlation between samples of the same population. I would guess running a chi square test would reveal a number that would reject the hypothesis that regional and major pilots are of the same population. I say this because the nature of the operations are different. Worldwide theatre vs east coast, mountainous terrain vs. flat, ocean crossings vs non, various fleet types vs single fleet type, number of seats, number of legs flown, stage length...and on and on and on. There are enough non-common variables to throw a wrench into a valid correlation.
Using an analysis of covariance with multiple regressions of the variables involved, you could determine a level of correlation between the groups. The differences and the initial status of the variables can be removed statistically...so that they can be compared to an initial baseline computed for each group. This process will provide numbers called "residuals"...or what remains of each group after the inequalites have been removed.
Your conclusion would have been a more valid had you compared a 1500 hour major captain with a 25,000 major captain. Or a 1500 hour regional captain with a 25,000 regional captain. This would eliminate many intervening and extraneous variables. I would venture to hypothesize that 1500 hour captains in a 747 would produce a much more unfavorable accident rate than what you quoted for the majors above. The reason...the environment is much more complex....i.e. a low correlation between the environment of a major network vs a regional network. The characteristics of the population must be more common before you can make sense of the numbers you published above.
You are using some "USA Today" info gathering and statistical analysis above. I could use the same analysis to prove that blue uniforms are safer than black ones...and that blondes are safer than brunettes. This is the type of analysis that gives creedence to the fact that liars figure and figures lie.
What is the statistical significance of the facts? Using a statistical test such as a chi squared or mann whitney test use you can determine the level of correlation between samples of the same population. I would guess running a chi square test would reveal a number that would reject the hypothesis that regional and major pilots are of the same population. I say this because the nature of the operations are different. Worldwide theatre vs east coast, mountainous terrain vs. flat, ocean crossings vs non, various fleet types vs single fleet type, number of seats, number of legs flown, stage length...and on and on and on. There are enough non-common variables to throw a wrench into a valid correlation.
Using an analysis of covariance with multiple regressions of the variables involved, you could determine a level of correlation between the groups. The differences and the initial status of the variables can be removed statistically...so that they can be compared to an initial baseline computed for each group. This process will provide numbers called "residuals"...or what remains of each group after the inequalites have been removed.
Your conclusion would have been a more valid had you compared a 1500 hour major captain with a 25,000 major captain. Or a 1500 hour regional captain with a 25,000 regional captain. This would eliminate many intervening and extraneous variables. I would venture to hypothesize that 1500 hour captains in a 747 would produce a much more unfavorable accident rate than what you quoted for the majors above. The reason...the environment is much more complex....i.e. a low correlation between the environment of a major network vs a regional network. The characteristics of the population must be more common before you can make sense of the numbers you published above.
