Re: P-value from T-stat
- From: Ray Koopman <koopman@xxxxxx>
- Date: Mon, 5 Oct 2009 11:19:18 -0700 (PDT)
On Oct 5, 12:56 am, Graham Ashe <knight_arm...@xxxxxxxxx> wrote:
I plan to write a function in a computer program to perform a two sample t-test assuming unequal variances on any two samples of data passed to it. Unfortunately, all I can actually calculate is the t-statistic and DOF but not the p-value (which I need to know in order to determine if the difference in means is statistically significant or not; let's assume it's always a two-tailed test).
1. Is there a way to estimate the p-value from the t-statistic (without looking up a reference table); at least to know if the difference in means is significant?
2. On a side note, if through calculation I get a fractional value for the DOF (e.g. 7.85), am I supposed to round it to 8 or truncate it to 7?
If your program environment has an incomplete Beta function, often
denoted I_x(a,b), in its library then the 2-tailed p-value is I_x(a,b)
with x = df/(df + t^2), a = df/2, b = 1/2, where df = the degrees of
freedom, which need not be an integer.
.
- References:
- P-value from T-stat
- From: Graham Ashe
- P-value from T-stat
- Prev by Date: How do I handle these unsual stats?
- Next by Date: Re: How do I handle these unsual stats?
- Previous by thread: Re: P-value from T-stat
- Next by thread: Re: P-value from T-stat
- Index(es):
Relevant Pages
|
