Statistical programs usually provide either 3 or 4 digits. Sometimes
you will get a value of 0.0000 or a value of 0.0. In either case,
report the p-value as <0.0001. If you have a value of 0.0003 report
that as the p-value.
If you have a p of 0.9328 or 0.4152 then 2 significant digits are
accurate enough, ie, 0.93 or 0.42.
If you have a p of 0.0342 or 0.0142 then, again, use digits, ie, .034
or .014, as well as with a p-value of .0038 or .0011.
I believe the difference between 0.0000 and 0.0 is that the former is
rounded from a longer value, such as 0.000002, at the time of printing,
while the latter is numerically zero, at least to the accuracy of the
computer and software being used.
Regards,
David Smith
University of Texas School of Public Health
------------------------------
Date: Sun, 21 May 2006 16:24:54 -0400
From: Isaac Dialsingh <[log in to unmask]>
Subject: Re: Pvalue
Hello everyone,
It is a common outcome for reviewers of journals to request p-values up
= to the first non zero digit. For example, 0.000 might not be accepted
= since the last digit is 0 whereas 0.0002 is acceptable. How do you get
= the entire p-value in SPSS?
Isaac
|