selecting multiple cases

[William-85]

I am trying to compute an equation with mulitple variables summed multiple
times.  These variables are also undergoing log transformations in the
course of the equation.  Problem: a number of the starting values are zero
and the log transformation (rightly) doesn't like them.

I was wondering if there was a good way to write the syntax such that
those zero cases are skipped.

I've tried an IF command with "or" separating each variable name (i.e. If
(Var1>0) or (Var2>0) or ...) but this does not seem to do it.

It's probably worth noting that the syntax runs, just with numberous error
messages also popping up.  Unfortunatly, the basic equation is two bulky
(Smith and Wilson's E_Var test of Evenness if it matters) to sort of test
it by hand.

Any ideas?

[Melissa Ives]

You need to use 'and' in your if statement if you want to include only
records where every variable is >0.
or will include all variables as long as at least one of them is >0.

Melissa
The bubbling brook would lose its song if you removed the rocks.


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
William
Sent: Tuesday, April 03, 2007 1:48 AM
To: [hidden email]
Subject: [SPSSX-L] selecting multiple cases

I am trying to compute an equation with mulitple variables summed
multiple times.  These variables are also undergoing log transformations
in the course of the equation.  Problem: a number of the starting values
are zero and the log transformation (rightly) doesn't like them.

I was wondering if there was a good way to write the syntax such that
those zero cases are skipped.

I've tried an IF command with "or" separating each variable name (i.e.
If
(Var1>0) or (Var2>0) or ...) but this does not seem to do it.

It's probably worth noting that the syntax runs, just with numberous
error messages also popping up.  Unfortunatly, the basic equation is two
bulky (Smith and Wilson's E_Var test of Evenness if it matters) to sort
of test it by hand.

Any ideas?


PRIVILEGED AND CONFIDENTIAL INFORMATION
This transmittal and any attachments may contain PRIVILEGED AND
CONFIDENTIAL information and is intended only for the use of the
addressee. If you are not the designated recipient, or an employee
or agent authorized to deliver such transmittals to the designated
recipient, you are hereby notified that any dissemination,
copying or publication of this transmittal is strictly prohibited. If
you have received this transmittal in error, please notify us
immediately by replying to the sender and delete this copy from your
system. You may also call us at [phone] for assistance.

[Richard Ristow]

At 02:47 AM 4/3/2007, William wrote:

>Problem: a number of [my] starting values are zero and the log
>transformation (rightly) doesn't like them. I was wondering if there
>was a good way to write the syntax such that those zero cases are
>skipped.
>
>I've tried an IF command with "or" separating each variable name (i.e.
>If
>(Var1>0) or (Var2>0) or ...) but this does not seem to do it.

Do you want a SELECT IF? That should do it, with the logic you're
writing, if you want to drop cases having zero or negative values on
any of those variables.

There's always

SELECT IF MIN(VAR1 TO VAR...) GT 0.

CAREFUL: "SELECT IF" removes cases *permanently*. Make sure you have a
saved copy of your complete data file. (Do this, no matter what you
do.) You can use the TEMPORARY command before your SELECT IF,
calculations, and procedure.

Or, if you want to keep all the cases but have the log transforms come
out missing for values <=0, but don't want error messages, there is

IF  var1 GT 0  VAR1_LOG = LG10(VAR1).

Or,

MISSING VALUES var1 (LO THRU 0).
COMPUTE          VAR1_LOG = LG10(VAR1).

[Spousta Jan]

>Problem: a number of [my] starting values are zero and the log
>transformation (rightly) doesn't like them. I was wondering if there
>was a good way to write the syntax such that those zero cases are
>skipped.

My two cents:

1) Often a neater solution is to use ANY

if (~any(0, var1 to var20)) foo = ln(bar).

2) Sometimes it is possible to logarithm x + 1 instead of x (or more
generally to logarithm x + epsilon, epsilon > 0). With 1, zeros give
zeros while large values (> 100) are almost unchanged.

compute foo = ln(bar + 1).

Regards

Jan