Thursday, 22 October 2015

Getting a p-value for a T test in R (cumulative distribution functions)

Here's how to get a P-value for a t-test in R (I always forget how to do this!) This uses the cumulative distribution function.

Cumulative distribution function for a t-distributed variable
If your t-value is 0.1558572 and degrees of freedom is 22:
> 2*pt(-abs(t),df=n-1)
[1] 0.8776341
It doesn't matter if your t-value is negative.

Cumultative distribution function for a Normally distributed variable
We can also calculate it for a Normally distributed variable, eg. P(Z <= 0.9), where  Z is a standard Normal variable is:
> pnorm(0.9)
[1] 0.8159399

Cumulative distribution function for a Poisson distributed variable
eg. P(X > 11), where X is a Poisson(8) variable:
> ppois(11, lower.tail=FALSE)
[1] 0.111924

No comments: