View Single Post
Old January 12, 2009, 03:29 PM   #28
David Armstrong
Junior member
Join Date: January 24, 2005
Location: SW Louisiana
Posts: 2,289
One, the operative word in your response here, unfortunately for certainty, is "probably".
Of course, that is what predictive stats deal with, probability. They allow us to understand the dynamics of a situation. They let us know what will "probably" happen so we can build a response that deals with the probability of that response working. For example, in a murder situation, we know with a high degree of likelihood that going along with the killer's wishes is is probably going to result in you being killed, just like we know that in an armed robbery situation going along with the robber's wishes will probably minimize your loss and danger. You can never be certain, but you can know what tends to work out best and start from there.
But they cannot, by their nature, tell us anything predictive about a single individual or event.
Sure they do. I can predict that you will not live to be 300 years old. I can predict that in an NBA basketball game that each team will score at least one goal.
For example, I have a good friend who is alive today because she was sitting on her seatbelt during the accident. Had she been wearing it, she would not have survived.
That is an assumption, unless you have replicated the accident with her wearing the belt and she did die.
But it does help illustrate the limitations of statistical methods in individual prediction.
Of course there are limits, but that does not mean you cannot predict at all or without any level of certainty. There will always be some degree of uncertainty, but the lower one gets that the better.

Last edited by David Armstrong; January 12, 2009 at 03:44 PM.
David Armstrong is offline  
Page generated in 0.03341 seconds with 7 queries