Fwd: Re: Longitudinal ogistic regression

[SR Millis-3]

SR Millis <[hidden email]> wrote:  Date: Mon, 17 Jul 2006 14:27:31 -0700 (PDT)
From: SR Millis <[hidden email]>
Subject: Re: Longitudinal ogistic regression
To: Gene Maguin <[hidden email]>

  If subjects have multiple observations over time time, standard logistic regression is inappropriate. There a many texts now available on longitudinal data analysis: Hedeker & Gibbons (2006), Weiss (2005), Brown & Prescott (1999), and Singer & Willett (2003), Verbeke & Molenberghs (2000), and Fitzmaurice, Laird, & Ware (2004)---among others.

  You need to use either a mixed effects regression model for binary or ordinal outcomes -- or generalized estimating equations (GEE) models.

  They can be easily implemented in SAS, Stata, or S-Plus----I don't know SPSS's capability in this regard.

  SR Millis

Gene Maguin <[hidden email]> wrote:
  All,

I am analyzing some longitudinal data with dichotomous or ordinal variables.
I had thought to say

LOGISTIC REGRESSION T2 WITH G T1/ENTER G T1/ENTER G BY T1.

Or, for ordinal variables.

PLUM T2 BY G WITH T1/LOCATION INTERCEPT G T1
G BY T1/PRINT FIT PARAMETER TPARALLEL.

However, somebody here has commented that such analyses are incorrect but,
off the top of his head, couldn't recall the cite. Can anyone comment and,
if possible, give a cite. If this setup is incorrect, what are the
alternatives?

Thanks, Gene Maguin



Scott R Millis, PhD, MEd, ABPP (CN & RP)
Professor & Director of Research
Department of Physical Medicine & Rehabilitation
Wayne State University School of Medicine
261 Mack Blvd
Detroit, MI 48201
Email: [hidden email]
Tel: [phone]
Fax: [phone]

*********************************************************
This electronic message may contain information that is confidential and/or legally privileged. It is intended only for the use of the individual(s) and entity named as recipients in the message. If you are not an intended recipient of this message, please notify the sender immediately and delete the material from any computer. Do not deliver, distribute or copy this message, and do not disclose its contents or take any action in reliance on the information it contains. Thank you.


Scott R Millis, PhD, MEd, ABPP (CN & RP)
Professor & Director of Research
Department of Physical Medicine & Rehabilitation
Wayne State University School of Medicine
261 Mack Blvd
Detroit, MI 48201
Email: [hidden email]
Tel: [phone]
Fax: [phone]

*********************************************************
This electronic message may contain information that is confidential and/or legally privileged. It is intended only for the use of the individual(s) and entity named as recipients in the message. If you are not an intended recipient of this message, please notify the sender immediately and delete the material from any computer. Do not deliver, distribute or copy this message, and do not disclose its contents or take any action in reliance on the information it contains. Thank you.

[Dale Glaser]

As far as I know SPSS is not equipped to handle, at least in the mixed model option, anything other than continuous outcomes...however, the HLM software has options for binomial/ordinal outcomes, with an option for the PQL estimator for the binary outcome......I may be wrong but I think MLwin uses the MQL estimator for binary/logistic models.........dale

SR Millis <[hidden email]> wrote:  SR Millis wrote: Date: Mon, 17 Jul 2006 14:27:31 -0700 (PDT)
From: SR Millis
Subject: Re: Longitudinal ogistic regression
To: Gene Maguin

If subjects have multiple observations over time time, standard logistic regression is inappropriate. There a many texts now available on longitudinal data analysis: Hedeker & Gibbons (2006), Weiss (2005), Brown & Prescott (1999), and Singer & Willett (2003), Verbeke & Molenberghs (2000), and Fitzmaurice, Laird, & Ware (2004)---among others.

You need to use either a mixed effects regression model for binary or ordinal outcomes -- or generalized estimating equations (GEE) models.

They can be easily implemented in SAS, Stata, or S-Plus----I don't know SPSS's capability in this regard.

SR Millis

Gene Maguin wrote:
All,

I am analyzing some longitudinal data with dichotomous or ordinal variables.
I had thought to say

LOGISTIC REGRESSION T2 WITH G T1/ENTER G T1/ENTER G BY T1.

Or, for ordinal variables.

PLUM T2 BY G WITH T1/LOCATION INTERCEPT G T1
G BY T1/PRINT FIT PARAMETER TPARALLEL.

However, somebody here has commented that such analyses are incorrect but,
off the top of his head, couldn't recall the cite. Can anyone comment and,
if possible, give a cite. If this setup is incorrect, what are the
alternatives?

Thanks, Gene Maguin



Scott R Millis, PhD, MEd, ABPP (CN & RP)
Professor & Director of Research
Department of Physical Medicine & Rehabilitation
Wayne State University School of Medicine
261 Mack Blvd
Detroit, MI 48201
Email: [hidden email]
Tel: [phone]
Fax: [phone]

*********************************************************
This electronic message may contain information that is confidential and/or legally privileged. It is intended only for the use of the individual(s) and entity named as recipients in the message. If you are not an intended recipient of this message, please notify the sender immediately and delete the material from any computer. Do not deliver, distribute or copy this message, and do not disclose its contents or take any action in reliance on the information it contains. Thank you.


Scott R Millis, PhD, MEd, ABPP (CN & RP)
Professor & Director of Research
Department of Physical Medicine & Rehabilitation
Wayne State University School of Medicine
261 Mack Blvd
Detroit, MI 48201
Email: [hidden email]
Tel: [phone]
Fax: [phone]

*********************************************************
This electronic message may contain information that is confidential and/or legally privileged. It is intended only for the use of the individual(s) and entity named as recipients in the message. If you are not an intended recipient of this message, please notify the sender immediately and delete the material from any computer. Do not deliver, distribute or copy this message, and do not disclose its contents or take any action in reliance on the information it contains. Thank you.



Dale Glaser, Ph.D.
Principal--Glaser Consulting
Lecturer--SDSU/USD/CSUSM/AIU
4003 Goldfinch St, Suite G
San Diego, CA 92103
phone: [phone]
fax: [phone]
email: [hidden email]
website: www.glaserconsult.com