There might be interesting questions to be asked concerning why you want
to do this "with printout piles the height of my leg". But to deal with
the problem as you specified it:
I rather doubt that SPSS contains anything of the sort you describe.
But SPSS used to have a "user" module, with which one could execute
one's own routine from within SPSS. If that capacity still exists, you
could (at least in principle) write C code to modify the Pearson output.
As to the modification itself: Years ago, Roald Buehler's P-STAT system
("P" for "Princeton") had a useful routine called BPRINT, designed
expressly for correlation matrices but useful in other situations also.
Coefficients whose absolute magnitude was less than a given threshold
(and the threshold was supplied by the user) would be "printed" as a
blank space (hence BPRINT, for Blank PRINTing), and only the values
greater than the threshold value would be printed. You might consider
writing a C routine to accept the correlation matrix from SPSS, express
the coefficients to no more than 3 decimal places (F8.3 format, say),
convert that to alphabetic format, and substitute a field of blank
spaces whenever the [absolute] value was <= a specifiable threshold.
Rather more effective visually, I think, than highlighting, or changing
fonts or colour; and no more work in the programming.
Of course, if you had access to a working copy of P-STAT ... but I
haven't heard anything about P-STAT for, oh, several decades, and
suspect it may be utterly defunct.
If your "printout piles" arise from constructing correlations on the
same set of variables for a number of different samples or
subpopulations, and you're looking for a way of winnowing down the
quantity of output to inspect, you might consider constructing a matrix
that reports (instead of a correlation coefficient) the number of such
sets of data for which that coefficient exceeds the threshold. You'd
still have to examine the individual matrices, but could do so in a
slightly more informed frame of mind.
On Sun, 1 May 2004, pH+ wrote:
> Hey guys, help out a student trying to crunch massive amounts of data.
>
> Under SPSS, when I do correlations (like the Pearson), it flags the
> correlations with two asterisks. However, I'm working with printout
> piles the height of my leg, and I need them flagged more
> conspicuously...say, the significants bolded, or underlined, or print
> out a darker background for the box or something.
>
> I've gone through the options boxes, and couldn't find anything. I
> really don't want to export to html and write a C program that does
> this for me. Can someone help? Say, maybe there's a option we specify
> under the syntax box, or something.
------------------------------------------------------------
Donald F. Burrill [EMAIL PROTECTED]
56 Sebbins Pond Drive, Bedford, NH 03110 (603) 626-0816
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
