On Mon, 15 Mar 2004 09:51:01 -0500, "RichardM" <[EMAIL PROTECTED]>
wrote:

> I am doing binomial tests and reporting probabilities and I need to find an
> effect size measure that would allow one to assess the strength of the
> relationship.  I know many people present z but since the probabilities have
> not been estimated using an approximation of the normal distribution, I
> thought it would be misleading to present the value of z along with a
> probability computed using the binomial distribution.
> 

State the actual proportions.  
And state where the power estimate comes from.

Use the 2x2  table and chi squared, when the Ns are big
enough to justify approximations.



-- 
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
 - I need a new job, after March 31.  Openings? -
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to