choice of reference category only changes coefficient with uncorrelated random intercept and slope

On 12 Sep 2017, at 23:08 , Fox, John <jfox at mcmaster.ca> wrote:

Dear David and Ben,

I haven't worked out the implications specifically, but even in a linear model fit by least-squares, with no constraints on the inter-coefficient correlations, the correlation between the coefficients is influenced by the choice of reference level for a factor. That suggests to me that constraining the correlation to zero would affect the coefficients.

As I said, this is far short of a proof, but the result seems intuitively plausible.

Best,
John

--------------------------------------
John Fox, Professor Emeritus
McMaster University
Hamilton, Ontario, Canada
Web: socserv.mcmaster.ca/jfox

-----Original Message-----
From: R-sig-mixed-models [mailto:r-sig-mixed-models-bounces at r-
project.org] On Behalf Of David Sidhu
Sent: Friday, September 8, 2017 8:19 PM
To: Ben Bolker <bbolker at gmail.com>
Cc: r-sig-mixed-models at r-project.org
Subject: Re: [R-sig-ME] choice of reference category only changes
coefficient with uncorrelated random intercept and slope

Hi Ben

Thanks for the reply.
Just to follow up, I tried running an lmer instead of a glmer and the
same thing happens: when a random slope and intercept are uncorrelated,
the choice of the reference category affects the absolutely value of
that predictor?s coefficient.

Dave

---
David M. Sidhu, MSc<http://davidmsidhu.com/> PhD Candidate Department of
Psychology University of Calgary






On Sep 8, 2017, at 12:04 PM, Ben Bolker
<bbolker at gmail.com<mailto:bbolker at gmail.com>> wrote:

Not sure, but ...

I think this is real. (If I were going to pursue it further I would
probably try running some simulations.)  I think the asymmetry you're
seeing is most likely related to the nonlinearity inherent in a GLMM; if
that's true, then the effect should go away if you were using a LMM
instead of a GLMM ...


On Tue, Sep 5, 2017 at 7:45 PM, David Sidhu
<dsidhu at ucalgary.ca<mailto:dsidhu at ucalgary.ca>> wrote:

Hi Everyone

I have noticed something strange...

I am running a glmer with a single dichotomous predictor (coded 1/0).
The model also includes a random subject intercept, as well as a random
item intercept and slope.

Changing which level of the predictor serves as the reference category
doesn?t change the absolute value of the coefficient, EXCEPT when the
random intercept and slope are uncorrelated.

This happens whether I keep the predictor as a numeric variable, or
change the predictor into a factor and use the following code:

t1<-glmer(DV~IV+(1|PPT)+(0+dummy(IV, "1")|Item)+(1|Item), data = data,
family = "binomial?)

Is this a genuine result? If so, can anyone explain why the uncorrelated
random intercept and slope allow it to emerge? If not, how can I run a
model that has an uncorrelated random intercept and slope that would
prevent the choice of reference category from affecting the result?

Thank you very much!

Dave

---
David M. Sidhu, MSc<http://davidmsidhu.com/> PhD Candidate Department of
Psychology University of Calgary







      [[alternative HTML version deleted]]