Posted by
Kylie on
Apr 07, 2011; 12:16am
URL: http://spssx-discussion.165.s1.nabble.com/Model-II-regression-analysis-tp4286435p4287499.html
Hi everyone,
The CNLR procedure can be used to fit major axis regression and reduced major
axis regression models in SPSS. Here are example syntax programs from the
knowledgebase (#12739, #12740).
=====================
There is not a built in procedure for reduced major axis regression in SPSS, but
this model can easily be fitted using the CNLR procedure. If Y is the dependent
variable and X is the predictor or independent variable, the command syntax
would be:
MODEL PROGRAM A=1 B=1.
COMPUTE PRED=A+B*X.
COMPUTE LOSS=((Y-PRED)**2)/ABS(B).
CNLR Y /LOSS=LOSS.
Bootstrapped estimates of standard errors are also available.
=====================
There is not a built in procedure for major axis regression
in SPSS, but this model can easily be fitted using the CNLR
procedure. If Y is the dependent variable and X is the
predictor or independent variable, the command syntax would
be:
MODEL PROGRAM A=1 B=1.
COMPUTE PRED=A+B*X.
COMPUTE LOSS=((Y-PRED)**2)/(1+B**2).
CNLR Y /LOSS=LOSS.
Bootstrapped estimates of standard errors are also available.
=====================
Hope this helps.
Cheers,
Kylie.
Quoting Mike Palij <
[hidden email]>:
> If the question is directed to me, I would have to say that I don't
> really know all that much about Model II regression (I'm more of
> a Model I kind of guy). However, there are some useful notes
> by Jack Weiss at UNC for one of his course that clarify the different
> types of models for error in variables in regression analysis; see:
>
http://www.unc.edu/courses/2007spring/biol/145/001/docs/lectures/Nov3.html#maxlikelihoodform> I assume that the OP's comment of reporting "ordinary least-products"
> may have something to do with "Alternative 3" which minimizes the
> product of the error in x times the error in y (also called geometric mean
> or reduced major axis regression).
>
> One of the key distinctions, the ratio of error in Y relative to error in X,
> is explored in more detail in another lecture; see:
>
http://www.unc.edu/courses/2007spring/biol/145/001/docs/lectures/Nov10.html>
> I presume that SEM can be used to estimate the parameters in the
> equation but it also seems likely to me that this type of analysis might
> be done by specialized software.
>
> I will now defer to someone who actually is familiar with this stuff.
>
> -Mike Palij
> New York University
>
[hidden email]
>
>
> ----- Original Message -----
> From: "Gene Maguin" <
[hidden email]>
> To: <
[hidden email]>
> Sent: Wednesday, April 06, 2011 2:10 PM
> Subject: Re: Model II regression analysis
>
>
> > Let me see if in can learn something here from this. I looked at the page
> on
> > the Monterey Bay website that you cited. From that description, it sounds
> as
> > though Model II regression is concerned with predictor variables measured
> > with error (which, I think, is referred to as the 'errors in variables'
> > problem). Error that could, perhaps, be estimated by specific procedures
> and
> > estimates of that error then used to correct a correlation/covariance
> matrix
> > in an SEM analysis. Would that be a correct line of thinking?
> >
> > Gene
> >
> >
> >
> >
> >
> >
> > -----Original Message-----
> > From: Mike Palij [mailto:
[hidden email]]
> > Sent: Wednesday, April 06, 2011 12:30 PM
> > To: Gene Maguin;
[hidden email]
> > Cc: Mike Palij
> > Subject: Re: Model II regression analysis
> >
> > I'm not speaking for Hazim but "Model II" regression is a specialized
> > type of regression analysis. Sokal and Rohlf present and contrast
> > Model I (ordinary regression for prediction) and Model II (major axis
> > regression) in their Biometry textbook, sections of which are available
> > at books.google.com; see:
> >
http://tinyurl.com/sokal-rohlf> > or
> >
>
http://books.google.com/books?id=N6KCNw5NHNkC&pg=PA543&dq=%22model+II%22+reg> >
> ression+%22major+axis%22&hl=en&ei=cpKcTYTxIsSdgQeltZiZBw&sa=X&oi=book_result
> >
> &ct=result&resnum=1&ved=0CCkQ6AEwAA#v=onepage&q=%22model%20II%22%20regressio
> > n%20%22major%20axis%22&f=false
> >
> > A brief historical presentation on the distinction is also available here:
> >
http://www.mbari.org/staff/etp3/regress/history.htm> >
> > There are several programs and routines that use different methods
> > for Model II regression (stand alone and programs for MatLab and R)
> > but I haven't seen "ordinary least products" used.
> >
> > -Mike Palij
> > New York University
> >
[hidden email]
> >
> >
> > ----- Original Message -----
> > From: "Gene Maguin" <
[hidden email]>
> > To: <
[hidden email]>
> > Sent: Wednesday, April 06, 2011 11:58 AM
> > Subject: Re: Model II regression analysis
> >
> >
> >> Hazim,
> >>
> >> I'm sure others on the list will know but what specifically is the
> >> design/meaning of an "ordinary least products"/Model II regression?
> >>
> >> Gene Maguin
> >>
> >>
> >>
> >> -----Original Message-----
> >> From: SPSSX(r) Discussion [mailto:
[hidden email]] On Behalf Of
> >> Hazim Markos
> >> Sent: Wednesday, April 06, 2011 10:43 AM
> >> To:
[hidden email]
> >> Subject: Model II regression analysis
> >>
> >> Hi,
> >> I'm a physiologist so here comes a cry for help. I just got a paper back
> >> from a journal editor who prefers I use "ordinary least products"
> > regression
> >> analysis, which I've since found out is also called Model II regression
> >> analysis (right?). I dont want to argue with him, so I've agreed to do
> > this.
> >>
> >> Long story short: how do I do this type of regression in SPSS v.17???
> >>
> >> Thanks
> >> Hazim
> >>
> >> =====================
> >> 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
>
> =====================
> 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
>
--
Kylie Lange
Biostatistician
Centre of Clinical Research Excellence (CCRE) in
Nutritional Physiology, Interventions and Outcomes
Discipline of Medicine
The University of Adelaide
Phone (Mon-Wed, Fri @ UniAdel): (08) 8222 5973
Phone (Thurs @ CSIRO): (08) 8303 8860
Fax: (08) 8223 3870
Web: www.adelaide.edu.au/ccre-nutrition
E-mail:
[hidden email]
CRICOS Provider Number 00123M
-----------------------------
IMPORTANT: This message may contain confidential or legally privileged
information. If you think it was sent to you by mistake, please delete all
copies and advise the sender. For the purposes of the SPAM Act 2003, this email
is authorised by The University of Adelaide.
Think green: read on the screen.
=====================
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