|WikiProject Statistics||(Rated Start-class, High-importance)|
I know the distinction between strict and non-strict inequality doesn't matter for CONTINUOUS random variables, but it DOES matter for discrete ones, and there are situations where it's useful to use the survival function for discrete variables, and you will screw up if you use non-strict.
I'm sorry, but I can only assume that this is wrong. If we were to generate a discrete survival function from a sample set (say, Bob, Jill, and Fred), we would have to take into account every time period that each were alive. So, in this case, lets say we use decade intervals to generate our survival function. Then, we have that Bob died when he was in his 20's, Jill died when she was in her 50's, and then Fred died when he was in his teens. f(0 to 10) = 3 because they were all alive then, but f(<10 to 20) = 2 because Fred died, then f(<20 to 50) = 1 because that's when Bod died, and then f(<50 to inf) = 0 because then Jill died. Note, we have to take into account all times each was alive, thus the function must be monotone decreasing. 22.214.171.124 (talk) 06:06, 8 June 2011 (UTC)SomeGuyWhoApparentlyKnowsSomeMath
left continuous vs. right continuous
The distribution function is right continuous, so survivor function is left continuous. Jackzhp (talk) 20:55, 28 January 2011 (UTC) Not true. The definition of right-continuous is preserved under the map f(x) |-> 1-f(x). — Preceding unsigned comment added by 126.96.36.199 (talk) 13:30, 16 May 2013 (UTC)