Reply
Tue 11 May, 2004 04:34 pm
The point of this discussion is for use in a research paper (not a very serious one), and I don't intend to prove my opinion, but merely express it. I point that out since my opinion has no real utility, and in fact is largely a debate of semantics, which few people enjoy. So keep in mind this discussion is for fun... with a purpose.
Percentages really only seem to apply in a limited number of cases. I have identified two, but there are probably more. One case is in describing the accuracy of a guess, rather than the guess itself. For example, if I say there is a 5% chance the world will blow up tomorrow, that cannot be true. Either the world will blow up or it will not, there is no middle ground. The probability of every event is either 1 or 0 (as someone mentioned on my "Randomness" thread), and the 5% only expresses percieved accuracy.
A second use for percentages is to measure controlled or static events. For example, past events are static, so one may accurately say that one in five people who were killed in an accident were hispanic.
On the other hand a serial killer who likes to kill 1 hispanic woman for every five of another race, would be a controlled event, and the percentages in both cases are legitimate. An example of when percentages would NOT be legitimate would be stating that one in five people die from cancer (forgive me for not using actual statistics). That statement cannot be true. You may say that one in five HAVE, but to say that one in five DO has multiple errors. I think I'll move onto a different point without covering myself on that last statement. Even if you do accept the latter wording as legitimate, you could not equate the same to mean that any one person has a 20% chance of dying from cancer. Because the chance of any specific person dying from lung cancer is 1 or 0. And the percentage only holds weight when describing a trend, for reasons already covered.
There, now someone come and rip me apart. It will be most useful.
Re: Use of Percentages
SCoates wrote:..........One case is in describing the accuracy of a guess, rather than the guess itself. For example, if I say there is a 5% chance the world will blow up tomorrow, that cannot be true. Either the world will blow up or it will not, there is no middle ground. The probability of every event is either 1 or 0 (as someone mentioned on my "Randomness" thread), and the 5% only expresses percieved accuracy.
wrong; what you are trying to express is the statistical 'probability' of some event. this is 'quantified' by applicable factors acting upon the event to render it more or less 'probable'. I agree that it is not what might be called a 'precise' science, however the numbers produced by a competent exponent of the practice are useful.
[and 'percentage has nothing to do with it; it is merely the chosen descriptive form.]
SCoates wrote:A second use for percentages is to measure controlled or static events. For example, past events are static, so one may accurately say that one in five people who were killed in an accident were hispanic.
On the other hand a serial killer who likes to kill 1 hispanic woman for every five of another race, would be a controlled event, and the percentages in both cases are legitimate. An example of when percentages would NOT be legitimate would be stating that one in five people die from cancer (forgive me for not using actual statistics). That statement cannot be true. You may say that one in five HAVE, but to say that one in five DO has multiple errors. I think I'll move onto a different point without covering myself on that last statement. Even if you do accept the latter wording as legitimate, you could not equate the same to mean that any one person has a 20% chance of dying from cancer. Because the chance of any specific person dying from lung cancer is 1 or 0. And the percentage only holds weight when describing a trend, for reasons already covered. ........
and again in your second 'case', most comments on the likely cause of death are extremely accurate, if meticulously prepared, as they refer to such a large sample (say the entire population of a country) that the accuracy is close to perfect - one in five people 'will' die of cancer; however, the prediction does not refer to a specific person, but to a calculated 'fact'.
I don't believe probability is ever anything other than 1 or 0. Anything else would be inaccurate. Probability is just makeshift math, for when we are too ignorant to discern whether an event will or will not happen. An event cannot be more or less possible. It is possible or impossible. The same applies with probablility.
I'm not saying there is no use to the calculations. On the news it says there is an 80% chance of rain tomorrow. That's good to know, but however extensive the calculations or accurate the facts used in those calculations, they are just used to approach the truth as closely as possible, and yet there is insufficient information availbale so past trends are relied upon. Still whether it rains is a set event--either it will or won't, and that 80% only describes the accuracy of the calculations, and not in fact the weather.
however, if you give the precipitation predictions for the coming year, over the 365 days one stands a chance (if one does one's homework) of being quite accurate.
there is a large chance it will rain this year, and a close (very close) to zero chance that it won't!
I suspect that rather than disagreeing with me, I just stated my point of view very poorly. I can find plenty of flaws with the way I worded my points (obviously I was too lazy to even look up a real statistic, or even spellcheck...), but I think the essense of my confusion can be summed up neatly: there are not varying levels of "possible." So probability cannot measure possibility. Therefor it must be measuring something else. It is a form of accuracy, therefor it seems to me to measure the accuracy of calculations in regards to possibility, rather than possibility itself. Now, what flaws are there in that statement?
SCoates wrote:......there are not varying levels of "possible." So probability cannot measure possibility. Therefor it must be measuring something else. It is a form of accuracy, therefor it seems to me to measure the accuracy of calculations in regards to possibility, rather than possibility itself. Now, what flaws are there in that statement?
that seems to work; the 'possible', and the 'probable' are, of course, two very different things.
are you, perhaps, seeking to examine the nature of "chance"?
I already have. That sort of lead to this question. The more I consider "chance" or the term "random" the more it seems to be merely ignorance in disguise. Did you take part in the "randomness" thread I started? I guess I'll go check myself. I suppose if you support truly random events (by my understanding of the word) then debate could come down to just that issue.
SCoates wrote:......... The more I consider "chance" or the term "random" the more it seems to be merely ignorance in disguise. Did you take part in the "randomness" thread I started?....... I suppose if you support truly random events (by my understanding of the word) then debate could come down to just that issue.
haven't seen it (link?);
since in my universe 'everything occurs by 'chance' this could prove interesting!
(sorry for the 'temporal' lag there)
Hmm... I'm having trouble deleting the diplicate.
SCoates wrote:I don't believe probability is ever anything other than 1 or 0. Anything else would be inaccurate. Probability is just makeshift math, for when we are too ignorant to discern whether an event will or will not happen. An event cannot be more or less possible. It is possible or impossible. The same applies with probablility.
I think you are missing a component here. Probability requires two factors. The first is you binary event - the event will happen or it won't. The second is the timing of that event.
We all know that we will die. The bigger question (and harder to calculate) is when we will each die. Can you say with absolute precision either way that you will either live (a "1") or die (a "0") this coming Sunday?
The percentage of probability factor is a guessitmate of how close the act of you dying this Sunday is. If you are in good health and not planning any excessivly risky adventures on Sunday you'd probably assume there is a 99.9999% chance you'll live. But you have no way of being absolutely positive (a "1") at this very moment.
I won't agree that the possibility of the event is anything but 0% or 100%. I have agreed that if probability is defined as our perception of the accuracy of calculations, then the probabilities are accurate. I can think of no other useful definition, since our calculations or ignorance do nothing to influence the event.
Quantum physics disagrees.
I think a point you are missing here is that the basis of a percentage is to detirmine how many "1"s you would get if the event was to take place 100 times. The 1-0 theory i'll admit is accurate to a degree, but the true purpose of a probability is to speculate and provide an "educated guess" as to whether a "1" or a "0" is going to come up. Again there is no way of proving the percentage probability theory without time travel, so we can all try to keep it simple i suppose.
Pondering, I did cover that usefullness in above posts.
Tobruk, care to elaborate?
Stumble It!
Tweet This
Bookmark on Delicious
Share on Facebook
Share on MySpace