Date:         Wed, 13 Aug 1997 19:37:33 GMT
Reply-To:     Richard F Ulrich <wpilib+@PITT.EDU>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From:         Richard F Ulrich <wpilib+@PITT.EDU>
Organization: University of Pittsburgh
Subject:      Re: significance of correlation coefficient

Richard Batt (rbatt@arts.adelaide.edu.au) wrote:
: Hi!
 
: For a simple correlation (N=35) SPSS 7 for Windows gives p = 0.05 for
: approx r = 0.330.  However when I check by hand using,
: p_val = 1 - CDF.T(t,33)   where t = r * sqrt( (33/(1-r^2) )
: I get p = 0.05 for approx 0.295.  Any suggestions as to why the
: discrepancy?  In my simulations I was doing multiple correlations,
: perhaps SPSS adjusts the p value for this?
 
 - If you are doing MULTIPLE CORRELATIONs then you are looking at
multiple Rs which are inherently two-tailed.  You have described
your first correlation in contradictory ways, when you say "simple"
and then "multiple".
 
The CDF gives you one tail, which yields a one-tailed p-value, as you
show above.
 
 
 
Rich Ulrich, biostatistician                wpilib+@pitt.edu
http://www.pitt.edu/~wpilib/index.html   Univ. of Pittsburgh