Continuity correction.
- From: "a.riva@UCL" <anto.ilmago@xxxxxxxxx>
- Date: Tue, 23 Oct 2007 13:53:04 -0000
Hi everyone.
I'm trying to calculate some statistics with Excel, using the
NORMSDIST function to normalize a discontinuous distribution.
NORMSDIST(z) returns the cumulative area below the distribution from
negative to positive values of z, so that NORMSDIST(z) with z<0 is
exactly the same as 1-NORMSDIST(Abs(z)).
When I calculate "z" I should apply a continuity correction, is it
right?
To do that, should I add or subtract 0.5?
z=(m-0.5-M)/s
or
z=(m+0.5-M)/s
Surfing the Net I found discordant opinions. Should I subtract 0.5
from z>0 and add 0.5 to z<0? That is, subtract 0.5 from Abs(z) and
then give back the original sign of z?
Thank you vaery much,
Antonio.
.
- Follow-Ups:
- Re: Continuity correction.
- From: Ken Butler
- Re: Continuity correction.
- From: David Jones
- Re: Continuity correction.
- Prev by Date: Re: Distribution of a vowel on the page
- Next by Date: Re: Continuity correction.
- Previous by thread: assess design profiles for experiment
- Next by thread: Re: Continuity correction.
- Index(es):