Generating within-person coefficients

classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|

Generating within-person coefficients

Oliver
Hi everyone,

I conducted a longitudinal study in which X and Y were measured once a day
for 10 consecutive days. As can be seen attached, my dataset is organized in
a "stacked" way so that I can conduct multilevel analyses.  I was wondering
whether a syntax could be used to generate the "within-person" coefficients
for the association between X and Y ? For instance, for each participant
(ID), I would like to know the magnitude of the association between X and Y.
Is there a way to generate either a correlation coefficient (or any other
statistical estimate quantifying the strength of the X-Y association),
separately for each ID ? I'm not sure if this would be a correlation
coefficient or a beta.

Thanks in advance for your support.
O.

<http://spssx-discussion.1045642.n5.nabble.com/file/t340718/Within-person_estimate.jpg>



--
Sent from: http://spssx-discussion.1045642.n5.nabble.com/

=====================
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
Reply | Threaded
Open this post in threaded view
|

Re: Generating within-person coefficients

Andy W
For exploratory data analysis, you can do something like:

*****************.
SORT CASES BY ID Wave.
SPLIT FILE BY ID.
REGRESSION
  /DEPENDENT y
  /METHOD=ENTER x.
SPLIT FILE OFF.
*****************.

Or swap out CORRELATIONS instead of regression.

For more complicated models (e.g. growth random-effect/fixed-effect), check
out the MIXED command and the SOLUTION option on the RANDOM subcommand
(available as of V25). Or you can do interaction effects with ID and x to
generate person level coefficients.



-----
Andy W
[hidden email]
http://andrewpwheeler.wordpress.com/
--
Sent from: http://spssx-discussion.1045642.n5.nabble.com/

=====================
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
Andy W
apwheele@gmail.com
http://andrewpwheeler.wordpress.com/
Reply | Threaded
Open this post in threaded view
|

Re: Generating within-person coefficients

Rich Ulrich
In reply to this post by Oliver
Andy suggested the simplest thing to implement.  I've done a couple of other
things, once or twice.

By the way, for the sake of generality, I want to mention that your example
does NOT show any strong trend of linearity across time, which is often an
artifact that confuses interpretations of "time-series" data.  (Hmm, a program
designed for time series might offer desirable options for handling these data.)

You can use Aggregate by ID to collect sums, sums of the squares and cross-products;
and in another step, compute statistics by person from the formulas  for r or b. 
I would also look briefly at whether the variances vary a lot across person, since a
tiny variance for x or y for a person would yield an r or b with a huge error band.
(The number of scores above 90, using a 0-100 range, makes me wonder if you
have a "ceiling effect" which might be deserve some compensatory transformation.
But that wasn't your question.)

Also, for a single Within-Subject correlation matrix across all subjects --
If the number of IDs is not too many, you can use Discriminant Function, where
ID is used as "groups"; that matrix is among the options for output.

--
Rich Ulrich


From: SPSSX(r) Discussion <[hidden email]> on behalf of Oliver <[hidden email]>
Sent: Monday, September 23, 2019 8:22 PM
To: [hidden email] <[hidden email]>
Subject: Generating within-person coefficients
 
Hi everyone,

I conducted a longitudinal study in which X and Y were measured once a day
for 10 consecutive days. As can be seen attached, my dataset is organized in
a "stacked" way so that I can conduct multilevel analyses.  I was wondering
whether a syntax could be used to generate the "within-person" coefficients
for the association between X and Y ? For instance, for each participant
(ID), I would like to know the magnitude of the association between X and Y.
Is there a way to generate either a correlation coefficient (or any other
statistical estimate quantifying the strength of the X-Y association),
separately for each ID ? I'm not sure if this would be a correlation
coefficient or a beta.

Thanks in advance for your support.
O.

<http://spssx-discussion.1045642.n5.nabble.com/file/t340718/Within-person_estimate.jpg>



--
Sent from: http://spssx-discussion.1045642.n5.nabble.com/

=====================
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
Reply | Threaded
Open this post in threaded view
|

Re: Generating within-person coefficients

Bruce Weaver
Administrator
In reply to this post by Andy W
Re Andy's first suggestion, you could use OMS to write the table of
coefficients from the regression model to another dataset for later use.
You'd also have the SEs of the coefficients, which could come in handy if
you wanted to do inverse-variance weighting (for example) to deal with the
issue Rich mentioned about variation in how precisely the slopes are
estimated.  

HTH.


Help for OMS:

https://www.ibm.com/support/knowledgecenter/en/SSLVMB_26.0.0/statistics_reference_project_ddita/spss/base/syn_oms.html




Andy W wrote

> For exploratory data analysis, you can do something like:
>
> *****************.
> SORT CASES BY ID Wave.
> SPLIT FILE BY ID.
> REGRESSION
>   /DEPENDENT y
>   /METHOD=ENTER x.
> SPLIT FILE OFF.
> *****************.
>
> Or swap out CORRELATIONS instead of regression.
>
> For more complicated models (e.g. growth random-effect/fixed-effect),
> check
> out the MIXED command and the SOLUTION option on the RANDOM subcommand
> (available as of V25). Or you can do interaction effects with ID and x to
> generate person level coefficients.
>
>
>
> -----
> Andy W

> apwheele@

> http://andrewpwheeler.wordpress.com/
> --
> Sent from: http://spssx-discussion.1045642.n5.nabble.com/
>
> =====================
> To manage your subscription to SPSSX-L, send a message to

> LISTSERV@.UGA

>  (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





-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
Sent from: http://spssx-discussion.1045642.n5.nabble.com/

=====================
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
--
Bruce Weaver
bweaver@lakeheadu.ca
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

PLEASE NOTE THE FOLLOWING: 
1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above.
2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/).
Reply | Threaded
Open this post in threaded view
|

Re: Generating within-person coefficients

Jon Peck
If you want to do predictions using results from REGRESSION or many other procedures, the best way is to save the model (Linear > Save > Export model information to XML file) and then use Utilities > Scoring Wizardd to apply it to new data.  That handles things like variable renaming, missing value treatment, and standard errors.

On Tue, Sep 24, 2019 at 12:43 PM Bruce Weaver <[hidden email]> wrote:
Re Andy's first suggestion, you could use OMS to write the table of
coefficients from the regression model to another dataset for later use.
You'd also have the SEs of the coefficients, which could come in handy if
you wanted to do inverse-variance weighting (for example) to deal with the
issue Rich mentioned about variation in how precisely the slopes are
estimated. 

HTH.


Help for OMS:

https://www.ibm.com/support/knowledgecenter/en/SSLVMB_26.0.0/statistics_reference_project_ddita/spss/base/syn_oms.html




Andy W wrote
> For exploratory data analysis, you can do something like:
>
> *****************.
> SORT CASES BY ID Wave.
> SPLIT FILE BY ID.
> REGRESSION
>   /DEPENDENT y
>   /METHOD=ENTER x.
> SPLIT FILE OFF.
> *****************.
>
> Or swap out CORRELATIONS instead of regression.
>
> For more complicated models (e.g. growth random-effect/fixed-effect),
> check
> out the MIXED command and the SOLUTION option on the RANDOM subcommand
> (available as of V25). Or you can do interaction effects with ID and x to
> generate person level coefficients.
>
>
>
> -----
> Andy W

> apwheele@

> http://andrewpwheeler.wordpress.com/
> --
> Sent from: http://spssx-discussion.1045642.n5.nabble.com/
>
> =====================
> To manage your subscription to SPSSX-L, send a message to

> LISTSERV@.UGA

>  (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





-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
Sent from: http://spssx-discussion.1045642.n5.nabble.com/

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


--
Jon K Peck
[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
Reply | Threaded
Open this post in threaded view
|

Re: Generating within-person coefficients

Oliver
In reply to this post by Oliver
Hi everyone,

Thanks for getting back to me with suggestions. Everything worked !

Thanks again.
MO.



--
Sent from: http://spssx-discussion.1045642.n5.nabble.com/

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