From: Giovanni on 4 Jun 2010 04:34 This is to be just a note about probability functions. When trying to compute the probability associated with some "noteworthy" probability functions (for instance, t-Student or Fisher) using the appropriate functions on my Hp 50g (so, for instance, UTPT or UTPF), the maximum number of degrees of freedom the calculator accepts will be 499. Is this normal? Is it happening only to me? I noticed that when trying to compute the p-value associated with some value of a Fisher probability function. It was about inference in a linear multi-variable model, with quite a big (975) number of observations. Is there a workaround for such a limitation? Beyond that, I cannot understand why it won't do such a calculation: when trying to do the same with Excel, the resulting number is bigger then the lowest number supported by the calculator (which I think is something*10^-499). Any contribution would be appreciated. Thanks! From: John H Meyers on 4 Jun 2010 07:24 "In probability and statistics, Student's t-distribution (or simply the t-distribution) is a continuous probability distribution that arises in the problem of estimating the mean of a normally distributed population when the sample size is small... As the number of degrees of freedom grows, the t-distribution approaches the normal distribution." http://en.wikipedia.org/wiki/Student's_t-distribution "As n increases, Student's t-distribution approaches the normal distribution." http://mathworld.wolfram.com/Studentst-Distribution.html I suppose that some of the intermediate calculations might also exceed the calculator's floating-point range (although there is a much higher range for internal extended precision, if occasional conversion to normal precision can be avoided for all intermediate results). [r->] [OFF] From: Giovanni on 4 Jun 2010 07:38 On 4 Giu, 13:24, John H Meyers wrote:> "In probability and statistics, > Student's t-distribution (or simply the t-distribution) > is a continuous probability distribution that arises > in the problem of estimating the mean > of a normally distributed population > when the sample size is small... > I was aware of this; but what about the F distribution? From: Giovanni on 4 Jun 2010 07:42 On 4 Giu, 13:24, John H Meyers wrote:> "In probability and statistics, > Student's t-distribution (or simply the t-distribution) > is a continuous probability distribution that arises > in the problem of estimating the mean > of a normally distributed population > when the sample size is small... > > As the number of degrees of freedom grows, > the t-distribution approaches the normal distribution."http://en.wikipedia.org/wiki/Student's_t-distribution > > "As n increases, > Student's t-distribution approaches the normal distribution."http://mathworld.wolfram.com/Studentst-Distribution.html > I was aware of this; what about the F distribution then? > I suppose that some of the intermediate calculations > might also exceed the calculator's floating-point range > (although there is a much higher range for internal > extended precision, if occasional conversion to normal precision > can be avoided for all intermediate results). > > [r->] [OFF] I haven't understood this part; could you be more specific, please? Obviously it's not a problem that of the probability functions; it's just a chance to learn more about this fantastic calculator! From what I understood, you can occasionally make more precise calculations than the calculator would normally let you do? How can you do that?  |  Pages: 1 Prev: HP50G: Emulator hangs when alarm goes offNext: USB TO IR TO HP48