Date: Wed, 30 Sep 1998 14:06:32 -0500
Reply-To: "Dr. Larry Grimm" <lgrimm@UIC.EDU>
Sender: "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From: "Dr. Larry Grimm" <lgrimm@UIC.EDU>
Subject: Re: p = .051
Content-Type: text/plain; charset="us-ascii"
As Jacob Cohen said, "surely God loves the .06 level as much as the .05
level." There is a voluminous literature on the issue, which amounts to
complaining about slavishly adhering to p values as the be all and end all.
What you can do is refer to the .06 (and other close values) as marginally
significant. HOWEVER, the focus should be on the effect size! A
nontrivial effect size that accompanies a p value of .06 is much better
than a trivial effect size with a p = .000001. Confidence intervals are
also recommended but are currently not used as often as effect size
indices. Hope this helps. Larry
At 01:38 PM 9/30/1998 -0400, you wrote:
> I am looking for a reference.
> On numerous occasions I have gotten p value between .05 and .06. I
>religiously accepted the null under these circumstances. I was wondering if
>there was any literature around addressing the issue of the .051 finding.
> Any references that you can suggest can be sent directly to me.
> TIA. Steve