Getting the Median for a Computed Variable

[Harding Faulk]

I am working with a professor on a hypotension study who has asked for
the calculation of the median on a computed variable.  Initially, there
were four variables whose coding response was 1-5 (i.e., strongly agree
to strongly disagree).  He wanted a combined variable called Total
Theoretical Motivation which was easily done using the compute function.
However, he then wanted to calculate the median on this combined
variable. Is this possible, since I have lost the individual scores when
I combined the four variables?  Is there another way to get the median
on a computed or aggregated variable?
 
Harding Faulk Jr.
Director, Institutional Research
Office of Institutional Research
Phone [phone]
Fax [phone]
[hidden email] <mailto:[hidden email]>
 
 
 

====================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

[ViAnn Beadle]

Median with respect to what? Per case, per group, or for all cases in the
file.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Harding Faulk
Sent: Thursday, October 18, 2007 4:10 PM
To: [hidden email]
Subject: Getting the Median for a Computed Variable

I am working with a professor on a hypotension study who has asked for
the calculation of the median on a computed variable.  Initially, there
were four variables whose coding response was 1-5 (i.e., strongly agree
to strongly disagree).  He wanted a combined variable called Total
Theoretical Motivation which was easily done using the compute function.
However, he then wanted to calculate the median on this combined
variable. Is this possible, since I have lost the individual scores when
I combined the four variables?  Is there another way to get the median
on a computed or aggregated variable?

Harding Faulk Jr.
Director, Institutional Research
Office of Institutional Research
Phone [phone]
Fax [phone]
[hidden email] <mailto:[hidden email]>




=======
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

[Art Kendall-2]

*<soapbox on>*
"since I have lost the individual scores when I combined the four variables? "

I hope that this is a typo!
Any analysis approach is subject to continuous improvement.  You need to
preserve all correct data in such a way that you can go back to the
beginning and refine the process. Having a complete "audit trail" mostly
benefits the analyst - you. This also means that your quality assurance
reviewer can go back and redo the analysis using variations on the
approach to see if they make a difference.  Likewise, anyone with whom
you share your data (as required by the ethics of various disciplines)
or who wants to adapt your approach to other data.

*<soapbox off>*

If you just want the median across all cases, just include that
specification on a FREQUENCIES command.

If you want to add a new variable to each case which is the median
across all cases, see the mode=ADDVARIABLES option in AGGREGATE.

If you mean something else, please give more details.

Art Kendall
Social Research Consultants

Harding Faulk wrote:

> I am working with a professor on a hypotension study who has asked for
> the calculation of the median on a computed variable.  Initially, there
> were four variables whose coding response was 1-5 (i.e., strongly agree
> to strongly disagree).  He wanted a combined variable called Total
> Theoretical Motivation which was easily done using the compute function.
> However, he then wanted to calculate the median on this combined
> variable. Is this possible, since I have lost the individual scores when
> I combined the four variables?  Is there another way to get the median
> on a computed or aggregated variable?
>
> Harding Faulk Jr.
> Director, Institutional Research
> Office of Institutional Research
> Phone [phone]
> Fax [phone]
> [hidden email] <mailto:[hidden email]>
>
>
>
>
> ===================
> 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

[ViAnn Beadle]

Run FREQUENCIES on that variable and ask for the median.

-----Original Message-----
From: Harding Faulk [mailto:[hidden email]]
Sent: Friday, October 19, 2007 7:07 AM
To: ViAnn Beadle
Subject: RE: Getting the Median for a Computed Variable - Reply

I need the median for that particular variable on all cases.

-----Original Message-----
From: ViAnn Beadle [mailto:[hidden email]]
Sent: Thursday, October 18, 2007 9:03 PM
To: Harding Faulk; [hidden email]
Subject: RE: Getting the Median for a Computed Variable

Median with respect to what? Per case, per group, or for all cases in
the
file.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Harding Faulk
Sent: Thursday, October 18, 2007 4:10 PM
To: [hidden email]
Subject: Getting the Median for a Computed Variable

I am working with a professor on a hypotension study who has asked for
the calculation of the median on a computed variable.  Initially, there
were four variables whose coding response was 1-5 (i.e., strongly agree
to strongly disagree).  He wanted a combined variable called Total
Theoretical Motivation which was easily done using the compute function.
However, he then wanted to calculate the median on this combined
variable. Is this possible, since I have lost the individual scores when
I combined the four variables?  Is there another way to get the median
on a computed or aggregated variable?

Harding Faulk Jr.
Director, Institutional Research
Office of Institutional Research
Phone [phone]
Fax [phone]
[hidden email] <mailto:[hidden email]>




=======
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

[ViAnn Beadle]

You never said how you computed your variable. Why don't you ask the
professor what he expects the median to be?

-----Original Message-----
From: Harding Faulk [mailto:[hidden email]]
Sent: Friday, October 19, 2007 9:37 AM
To: ViAnn Beadle
Subject: RE: Getting the Median for a Computed Variable - Reply-Response

Thanks ViAnn but that doesn't appear to work because by computing the
new variable it changes the rating scale of 1-5.  I am beginning to
think that the professor shouldn't try to get the median on the newly
computed variable, especially if he expects the medial to between 1-5.

-----Original Message-----
From: ViAnn Beadle [mailto:[hidden email]]
Sent: Friday, October 19, 2007 10:46 AM
To: Harding Faulk
Cc: [hidden email]
Subject: RE: Getting the Median for a Computed Variable - Reply

Run FREQUENCIES on that variable and ask for the median.

-----Original Message-----
From: Harding Faulk [mailto:[hidden email]]
Sent: Friday, October 19, 2007 7:07 AM
To: ViAnn Beadle
Subject: RE: Getting the Median for a Computed Variable - Reply

I need the median for that particular variable on all cases.

-----Original Message-----
From: ViAnn Beadle [mailto:[hidden email]]
Sent: Thursday, October 18, 2007 9:03 PM
To: Harding Faulk; [hidden email]
Subject: RE: Getting the Median for a Computed Variable

Median with respect to what? Per case, per group, or for all cases in
the
file.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Harding Faulk
Sent: Thursday, October 18, 2007 4:10 PM
To: [hidden email]
Subject: Getting the Median for a Computed Variable

I am working with a professor on a hypotension study who has asked for
the calculation of the median on a computed variable.  Initially, there
were four variables whose coding response was 1-5 (i.e., strongly agree
to strongly disagree).  He wanted a combined variable called Total
Theoretical Motivation which was easily done using the compute function.
However, he then wanted to calculate the median on this combined
variable. Is this possible, since I have lost the individual scores when
I combined the four variables?  Is there another way to get the median
on a computed or aggregated variable?

Harding Faulk Jr.
Director, Institutional Research
Office of Institutional Research
Phone [phone]
Fax [phone]
[hidden email] <mailto:[hidden email]>




=======
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

[Nicole Haag]

If you compute your new variable as the mean of the original 4 variables the scale should still be 1-5 and so should the median.

>You never said how you computed your variable. Why don't you ask the
>professor what he expects the median to be?
>
>-----Original Message-----
>From: Harding Faulk [mailto:[hidden email]]
>Sent: Friday, October 19, 2007 9:37 AM
>To: ViAnn Beadle
>Subject: RE: Getting the Median for a Computed Variable - Reply-Response
>
>Thanks ViAnn but that doesn't appear to work because by computing the
>new variable it changes the rating scale of 1-5.  I am beginning to
>think that the professor shouldn't try to get the median on the newly
>computed variable, especially if he expects the medial to between 1-5.
>
>-----Original Message-----
>From: ViAnn Beadle [mailto:[hidden email]]
>Sent: Friday, October 19, 2007 10:46 AM
>To: Harding Faulk
>Cc: [hidden email]
>Subject: RE: Getting the Median for a Computed Variable - Reply
>
>Run FREQUENCIES on that variable and ask for the median.
>
>-----Original Message-----
>From: Harding Faulk [mailto:[hidden email]]
>Sent: Friday, October 19, 2007 7:07 AM
>To: ViAnn Beadle
>Subject: RE: Getting the Median for a Computed Variable - Reply
>
>I need the median for that particular variable on all cases.
>
>-----Original Message-----
>From: ViAnn Beadle [mailto:[hidden email]]
>Sent: Thursday, October 18, 2007 9:03 PM
>To: Harding Faulk; [hidden email]
>Subject: RE: Getting the Median for a Computed Variable
>
>Median with respect to what? Per case, per group, or for all cases in
>the
>file.
>
>-----Original Message-----
>From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
>Harding Faulk
>Sent: Thursday, October 18, 2007 4:10 PM
>To: [hidden email]
>Subject: Getting the Median for a Computed Variable
>
>I am working with a professor on a hypotension study who has asked for
>the calculation of the median on a computed variable.  Initially, there
>were four variables whose coding response was 1-5 (i.e., strongly agree
>to strongly disagree).  He wanted a combined variable called Total
>Theoretical Motivation which was easily done using the compute function.
>However, he then wanted to calculate the median on this combined
>variable. Is this possible, since I have lost the individual scores when
>I combined the four variables?  Is there another way to get the median
>on a computed or aggregated variable?
>
>Harding Faulk Jr.
>Director, Institutional Research
>Office of Institutional Research
>Phone [phone]
>Fax [phone]
>[hidden email] <mailto:[hidden email]>
>
>
>
>
>=======
>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

=====================
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