It is statistically within the realm of possibility (meaning, better than 1 in a million chance of it NOT being a statistical fluke) that with the observed numbers (20 to 1), the observed difference is due to a real (on the ground, not observed) difference of NO LESS than 5 to 1
That is to say, the odds of it actually being 4 to 1 (not as…
It is statistically within the realm of possibility (meaning, better than 1 in a million chance of it NOT being a statistical fluke) that with the observed numbers (20 to 1), the observed difference is due to a real (on the ground, not observed) difference of NO LESS than 5 to 1
That is to say, the odds of it actually being 4 to 1 (not as bad) are like 1 in 100 million, while the odds of it being 10 to 1 (worse than 5 to 1) are like 1 in 10,000
If you were to take a 98% confidence interval (standard for peer review papers), which is 1 in 50 odds of the observed 20 to 1 difference being a statistical fluke, you would probably arrive at like 15 to 1
That's a much different way of looking at it - ie what are the odds that the odds would be a 20:1 ratio, tho I may still misunderstand. That is also why I learned on Day 1 in my class on Business Statistics that Statistics don't lie but Liars figure statistics. I am not saying that Steve is lying at all, but that anybody can use statistics to their advantage. It comes down to who has the best numbers, and that would clearly be Mr Kirsch.
Well we can never know whether what we observe is due to chance/how closely our observations of a small part of a population map to the whole population.
I don't know how Steve arrived at his figures, but I'm sure if he were to explain, then the method would be logically and mathematically sound.
If you were to shoot a shotgun into a flock of birds and you knew roughly the size of the flock, you could look at the ratio of dead red birds to green birds and construct a bell curve of possible ratios of red to green in the actual flock. If your ratio is 20 to 1 red to green, it becomes increasingly less likely that the actual ratio in the flock is the ratio (15, 12, 9, 5 etc to 1) as you move away from that number.
It is statistically within the realm of possibility (meaning, better than 1 in a million chance of it NOT being a statistical fluke) that with the observed numbers (20 to 1), the observed difference is due to a real (on the ground, not observed) difference of NO LESS than 5 to 1
That is to say, the odds of it actually being 4 to 1 (not as bad) are like 1 in 100 million, while the odds of it being 10 to 1 (worse than 5 to 1) are like 1 in 10,000
If you were to take a 98% confidence interval (standard for peer review papers), which is 1 in 50 odds of the observed 20 to 1 difference being a statistical fluke, you would probably arrive at like 15 to 1
That's a much different way of looking at it - ie what are the odds that the odds would be a 20:1 ratio, tho I may still misunderstand. That is also why I learned on Day 1 in my class on Business Statistics that Statistics don't lie but Liars figure statistics. I am not saying that Steve is lying at all, but that anybody can use statistics to their advantage. It comes down to who has the best numbers, and that would clearly be Mr Kirsch.
Well we can never know whether what we observe is due to chance/how closely our observations of a small part of a population map to the whole population.
I don't know how Steve arrived at his figures, but I'm sure if he were to explain, then the method would be logically and mathematically sound.
If you were to shoot a shotgun into a flock of birds and you knew roughly the size of the flock, you could look at the ratio of dead red birds to green birds and construct a bell curve of possible ratios of red to green in the actual flock. If your ratio is 20 to 1 red to green, it becomes increasingly less likely that the actual ratio in the flock is the ratio (15, 12, 9, 5 etc to 1) as you move away from that number.