Understanding/plotting fixed effects estimates &standard errors

I woke up this morning and realized I had given you bad advice.
changing the success/fail has no effect on the predictions, as you
found.

you need to put in alternate values of the estimated intercept.
Recall a logistic curve is S shaped. I suspect the intercept you are
using is placing you on the far left or right, where the effect of
change is not substantial. If you adjust the intercept estimate from
low to high, then you will see differences.

Think of this like a linear model, where the confidence interval is an
hour-glass shaped thing.  If you move from left to right, it shrinks
and grows.  Look on slide  37 in this lecture:
http://pj.freefaculty.org/guides/stat/Regression/ElementaryOLS/Regression-2-lecture.pdf
if you don't remember the hourglass

I believe similar is happening in your situation.  The difference is
that the left-and-right movement of the predicted value is driven by
your intercept.  The random effect affects the intercept, so in some
conditions the effect is larger than others, and the standard error
should also change.

Oops, I didn't even see that you presented the no-intercept results in your
first email. ?I would trust the LRTS results, I think. I would really need
to see the data to get a feel for what is happening (like what is the number
of trials etc). ?Do you even need variable?

--
| David Duffy (MBBS PhD) ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ,-_|\
| email: davidD at qimr.edu.au ?ph: INT+61+7+3362-0217 fax: -0101 ?/ ? ? *
| Epidemiology Unit, Queensland Institute of Medical Research ? \_,-._/
| 300 Herston Rd, Brisbane, Queensland 4029, Australia ?GPG 4D0B994A v

_______________________________________________
R-sig-mixed-models at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models

Paul E. Johnson
Professor, Political Science ? ?Assoc. Director
1541 Lilac Lane, Room 504 ? ? Center for Research Methods
University of Kansas ? ? ? ? ? ? ? University of Kansas
http://pj.freefaculty.org ? ? ? ? ? ?http://quant.ku.edu