p-values from tprob differ from those in excel or R

[rahulthathoo]

Hi.

I used Statisticsistribution:tprob to calculate the p-values in perl.
But the values that i got were exactly half the values i got for the
same parameters in Excel and R. How come? Any clues?

Rahul

 

[A. Sinan Unur]

I used Statisticsistribution:tprob to calculate the p-values in perl.
But the values that i got were exactly half the values i got for the
same parameters in Excel and R. How come? Any clues?

Do you know basic statistics? Statistics:istribution gives 1 - CDF.

On the other hand, the Excel function allows you to specify whether you
want a single tailed or two tailed p-value.

Syntax

TTEST(array1,array2,tails,type)

I am not very familiar with R, but it is probably the same deal.

See http://faculty.vassar.edu/lowry/t_single.html

Sinan

 

[xhoster]

Hi.

I used Statisticsistribution:tprob to calculate the p-values in perl.
But the values that i got were exactly half the values i got for the
same parameters in Excel and R. How come? Any clues?

It would help if you showed the exact syntax of all concerned. Excel has
several functions that deal with t distributions and they have different
behavior regarding cumulative/density and 1-tailed/2-tailed.

Statisticsistribution:tprob gives the upper probability, which if t is >
0 is the same thing as a one-tailed probability. You are probably asking
Excel for a two-tailed probability, which is double the one tail
probability.

Xho