Re: TDist Formula and Explanation

On Mar 28, 12:17 pm, Jack Tomsky <jtom...@xxxxxxxxxxxxx> wrote:
On Mar 28, 11:22 am, Jack Tomsky
<jtom...@xxxxxxxxxxxxx> wrote:
I am trying to derive the formula that MS Excel
uses
as part of the
tdist function to come up with the p-value. Can
anyone help me create
this formula so that I can calculate the p-value
for
one sample of
data?

Excel's tdist is not the cdf of t, but rather one
minus the cdf of the absolute value of t. So if the
argument t is negative, you would get an error
message. If the argument is positive, you get the
right-hand tail.

Suippose t is negative and you want the probability
that T < t. This probability can be calculated by
tdist(-t,m,1).

I find this annoying and can't understand why they
haven't fixed this flaw.

Jack

I've read that too, but I'm try to calculate the
value without using
any of Excel's blackbox formulas. Can anyone provide
me with some
guidance on that?

Abramowitz and Stegun has series expansions for the cdf of |t| on page 948. There's also an analytic result in terms of the incomplete beta function. If you don't have access to their book, I could reproduce their formulas.

Jack

Here is a URL to A&S.

http://www.math.sfu.ca/~cbm/aands/


.

Relevant Pages

  • Re: TDist Formula and Explanation
    ... tdist function to come up with the p-value. ... Excel's tdist is not the cdf of t, ... Suippose t is negative and you want the probability ...
    (sci.stat.math)
  • Re: TDist Formula and Explanation
    ... tdist function to come up with the p-value. ... Excel's tdist is not the cdf of t, but rather one minus the cdf of the absolute value of t. ... Suippose t is negative and you want the probability that T < t. ...
    (sci.stat.math)
  • Re: TDist Formula and Explanation
    ... tdist function to come up with the p-value. ... Excel's tdist is not the cdf of t, but rather one minus the cdf of the absolute value of t. ... Suippose t is negative and you want the probability that T < t. ...
    (sci.stat.math)
  • Re: How to compute p-value
    ... get the corresponding value of spot. ... > Presumably you have the Statistics Toolbox, since you're using CDF. ... > linear algebra function). ... > Neither of these two things is a p-value, ...
    (comp.soft-sys.matlab)