powerStudent

Approximate power estimation of a performed Student's t test about mean(s).

Description

Estimates the approximate statistical power of a performed Student's t test about mean(s). Here, according to Zar (1999) we use the approximation of power by the central cumulative distribution function. It recalls you the statistical result of the test you should have arrived. This power estimation is considered biased, for it is underestimated. However, it is presented for all those interested people.

You can find the exact power estimation on the extpowerStudent m-file.

Inputs:
t - Student's t statistic
df - degrees of freedom
c - specified testing direction [1 = one-tailed(default);2 = two-tailed]
alpha - significance level (default = 0.05)

Outputs:
- Specified direction test
(Statistical result of the test you should have arrived)
- Power

• Required Products: Statistics Toolbox
• MATLAB release: MATLAB 5.3 (R11)

[florabel cuaya]

 

[Linda Seltzer]

It seems that the power is being computed based on the t cumulative distribution function. However, if the null hypothesis is false, isn't the distribution noncentral, with the need to compute a noncentrality parameter? This means that operating curves have to be computed. Is there a Matlab routine somewhere that computes operating curves

[auto promotion]

Do you think others don't see what you are doing?

[Brahim HAMADICHAREF]

function [Power] = powerStudent(t,df,c,a)
rather than
function [powerStudent] = powerStudent(t,df,c,a)