Hi Lucia,
The third statistic in each cell of the Correlations table in Partial Corr output is the degrees of freedom (df), not the sample size. From the algorithms for Partial Corr (in Help->Algorithms), the degrees of freedom are calculated as:
df=M−theta−2
where theta is the order of the coefficient and M is the minimum sum of weights from which the zero-order coefficients involved in the computations were calculated. There is no weight variable assigned and no missing data in the 3 variables that you analyzed from the bankloan.sav file, so M equals the sample size, 850. There is one controlling variable in your analysis, so theta equals 1 and
df = 850 - 1 - 2 = 847 , which is the df reported in the partial correlation cells when I run the analysis that you describe. The df in the zero-order section (no controlling variables) is 850 - 2 = 848.
David Matheson
SPSS Statistical Support
-----Original Message-----
From: SPSSX(r) Discussion [mailto:
[hidden email]] On Behalf Of Lucia Costanzo
Sent: Monday, March 02, 2009 2:57 PM
To:
[hidden email]
Subject: partial correlation missing observations
Hi
I was wondering if anyone else is encountering this issue, when I run a partial correlation using the bankloan.sav file that comes with SPSS 16.0, setting the variables as debtinc and employ and the controlling variable as income. The analysis is missing 3 observations.
Any Ideas?
Thanks,
Lucia
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
Locked
