Numbers don't serve agendas, numbers exist. Simple uncomplicated facts. I think you mean that people with agendas use or misuse numbers. Or not, when they might prove more than a little embarrassing.Originally Posted by j2k4
I'm sorry, lynx, but your very able and astute explanation does not relieve my concern, nor does it mitigate my point, which is that numbers are constantly fudged, and people like yourself, in the face of being called for bullshit, then delve into statistical minutiae so as to obfuscate the larger issue of numbers being inflated or diminished to suit a preconceived notion or political agenda.What it means is that 95% of the possible results are expected to fall in the range 8.000-194,000. In other words, if you got another piece of data there is a 95% probability that the data would cause the likely total number to fall in the range 8,000-194,000. That may sound like a subtle difference, but it is actually very significant.
If one were simply counting numbers to reach the probable total, then another "piece of data" would be another dead body, and it would automatically shift the total higher. However, the people who compiled the report had to take into account how likely it was that a "body" was caused by the military intervention. Under those conditions another body which you could definitely say was caused by military intervention would cause the likely total to increase, while a body which you could definitely say was not caused by military intervention would cause the likely total to decrease.
Because of that uncertainty there is a wide spread of likely results. However, that spread will not be a flat line but is more likely to look something like this:
At you can see the likelihood of a central result is much higher that one at the fringes.
The significance of these figures becomes apparent not when you incorrectly snear at the 95%CI as Fred Kaplan did, but when you look at the likelihood that the total falls outside the range. It is not only a 95% confidence that the true figure falls inside that range, it is also a 95% confidence that it does not fall outside that range, both above and below, and there is actually a 97.5% confidence that the total is not below 8,000.
Would you place your money on that 2.5% chance of the true total being below 8,000?
Where is your proof that numbers are constantly fudged, or is this another of your unsupported generalities? Is it only numbers which contradict your viewpoint which are fudged?
Did I delve into statistical minutae, or was it your correspondent who quoted one piece of information (which on it's own is virtually meaningless) and tried to use it to denounce what must have been several man-years of effort? That's where we see bullshit coming in to play, and it was introduced by yourself.
My point is that this happens constantly, and that accurate numbers are next to impossible to come by, and your point is that certain numbers are de facto accurate, but one must have a mind-set similar to your own to know just which ones they are.
Perhaps you should go back and re-read my posts, you will find that I haven't actually claimed that the numbers are accurate. What I have pointed out (repeatedly) is that the people you have picked to denigrate the figures have done so with absolutely no basis whatsoever. Isn't it strange how they automatically know which ones are fudged.
It now occurs that experience with statistics aids greatly in giving one a leg-up in distilling these numbers and declaring their legitimacy for the benefit of the benighted.
Pity though, my point will not be addressed by such as you, capable as you are; a background in statistical analysis doesn't keep you from avoiding the issue I've raised.
If you present correspondents who quote statistics, you must expect statistics to be used to counter their argument. If for some reason you now think statistics shouldn't be used, then don't introduce them.
Numbers serve agendas, and no amount of testimonial as to the painstaking effort expended in the gathering of them changes the fact.
Reply With Quote
Bookmarks