SPSS regression coefficients and effect size

[Marsha and Mike SZYMCZUK]

A co-worker asked me to help her with her dissertation using SPSS regression.  

Her dissertation chair is asking her to compute effect sizes on the significant regression coefficients.  This is not a standard SPSS option.

She has both categorical and continuous variables.

Any ideas?

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

[Jason Burke]

What is the measurement level of her dependent variable?

Jason

On Fri, Sep 26, 2008 at 11:36 AM, Marsha and Mike SZYMCZUK
<[hidden email]> wrote:

> A co-worker asked me to help her with her dissertation using SPSS regression.
>
> Her dissertation chair is asking her to compute effect sizes on the significant regression coefficients.  This is not a standard SPSS option.
>
> She has both categorical and continuous variables.
>
> Any ideas?
>
> 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

[Garry Gelade]

Marsha & Mike

It depends what her chair means (or think he means) by 'effect size'. One
thing you can easily do in SPSS is to compute the change in r-squared when
an IV is added to the regression equation - on the regression dialogue box,
put each IV in a separate block, click Statistics and tick the R-squred
change box.

However, the answers you get will depend on the order the variables are
entered into the equation. One measure of effect size therefore is to find
the change in R-squared when the predictor is added LAST. So if you had 3
predictors, X,Y and Z, the effect size for Z would be the change in R-sq
when Z is added to a regression model containing X and Y; the effect size
for X would be the change in R-sq when X is added to a regression model
containing Y and Z, and so on.

If you have mutliple dummy variables representing a single category, I would
treat the dummies as a single block.

Garry Gelade

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Marsha and Mike SZYMCZUK
Sent: 26 September 2008 02:37
To: [hidden email]
Subject: SPSS regression coefficients and effect size

A co-worker asked me to help her with her dissertation using SPSS
regression.

Her dissertation chair is asking her to compute effect sizes on the
significant regression coefficients.  This is not a standard SPSS option.

She has both categorical and continuous variables.

Any ideas?

=======
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

__________ NOD32 3472 (20080925) Information __________

This message was checked by NOD32 antivirus system.
http://www.eset.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

[Melissa Ives]

Here's how we do it for various types of measures:

The effect sizes reported here are based on formula 1 for dichotomous measures, formula 2 for continuous measures, and formula 3 for chi-square conversion of survival analyses.
(1)     d= 2*arcsine (% from group2) - 2*arcsine (% from group1)
(2)     d= (group2 mean - group1 mean) / pooled standard deviation
(3)     d= square root ((4*(Chi-square))/(n-(Chi-square)))

See also: http://www.chestnut.org/LI/downloads/ESWK.xls for an excel worksheet.

Melissa

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Marsha and Mike SZYMCZUK
Sent: Thursday, September 25, 2008 8:37 PM
To: [hidden email]
Subject: [SPSSX-L] SPSS regression coefficients and effect size

A co-worker asked me to help her with her dissertation using SPSS regression.

Her dissertation chair is asking her to compute effect sizes on the significant regression coefficients.  This is not a standard SPSS option.

She has both categorical and continuous variables.

Any ideas?

=======
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

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.

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

[Henrik Lolle]

In the UNIANOVA procedure you can specify the option ETASQ which (in
the point and click user surface is called 'effect size'). This measure
is estimated for each variable (factors as well as covariates). Partial
ETA (not squared) compares to the partial correlation measure in
ordinary linear regression. There is, however, some discussion
concerning this measure as an indication of effect size. Alternatively,
you can via the old ANOVA procedure with the MCA option estimate ETAs
and BETAs, and these compare to bivariate and controlled BETAs in
ordinary linear regression.

Best,
Henrik

Quoting Melissa Ives <[hidden email]>:

> Here's how we do it for various types of measures:
>
> The effect sizes reported here are based on formula 1 for dichotomous
> measures, formula 2 for continuous measures, and formula 3 for
> chi-square conversion of survival analyses.
> (1)     d= 2*arcsine (% from group2) - 2*arcsine (% from group1)
> (2)     d= (group2 mean - group1 mean) / pooled standard deviation
> (3)     d= square root ((4*(Chi-square))/(n-(Chi-square)))
>
> See also: http://www.chestnut.org/LI/downloads/ESWK.xls for an excel
> worksheet.
>
> Melissa
>
> -----Original Message-----
> From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf
> Of Marsha and Mike SZYMCZUK
> Sent: Thursday, September 25, 2008 8:37 PM
> To: [hidden email]
> Subject: [SPSSX-L] SPSS regression coefficients and effect size
>
> A co-worker asked me to help her with her dissertation using SPSS regression.
>
> Her dissertation chair is asking her to compute effect sizes on the
> significant regression coefficients.  This is not a standard SPSS
> option.
>
> She has both categorical and continuous variables.
>
> Any ideas?
>
> =======
> 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
>
> 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.
>
> =====================
> 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
>

************************************************************
Henrik Lolle
Department of Economics, Politics and Public Administration
Aalborg University
Fibigerstraede 1
9200 Aalborg
Phone: [phone]
************************************************************

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