## Re: [Synoptic-L] When is a correlation meaningful?

Expand Messages
• Emmanuel Fritsch @abraxa.ign.fr on 01/30/2002 04:05:46 AM Sent by: fritsch@abraxa.ign.fr To: dgentil@sears.com cc:
Message 1 of 3 , Jan 30, 2002
Emmanuel Fritsch <emmanuel.fritsch@...>@... on 01/30/2002
04:05:46 AM

Sent by: fritsch@...

To: dgentil@...
cc: Synoptic-L@...

Subject: Re: [Synoptic-L] When is a correlation meaningful?

> If I were trying to construct some sort of forecast of one variable,
> based on the other, .7 might be a reasonable cut off. Half of the
> variation in one variable is explained by the other, at this level.
> ( .5^2 = appox. .7 )

.5^2=.25 (exactly) and .7^2=.5 (approx.)

*****Whoops. Thanks. I'll blame dyslexia for that. :o)

> However, our question is related to demonstrating a relationship,
> positive or negative, exists. So for our purposes, any correlation
> with a sufficiently large p value should be accepted.
> [...]
> Of course the correlations, even once accepted, could have more than
> one possible explanation. My opinion is that the major effects are
> related directly to authorship, but that there are other effects as
> well, artificially raising the confidence level might be one way to
> eliminate these, but it also makes the cut-off somewhat subjective,
> since we are now judging what results need to be excluded.

Some minor effects may be explaned from a composite authorship :
if two categories are correlated with r=0.15 (and p < .000056), it
may result from both categories sharing a common author, but only
on a small proportion of their extent.

a+
manu

******Agreed.

Dave Gentile
Riverside, Illinois
M.S. Physics
Ph.D. Management Science candidate

Synoptic-L Homepage: http://www.bham.ac.uk/theology/synoptic-l
List Owner: Synoptic-L-Owner@...
Your message has been successfully submitted and would be delivered to recipients shortly.