[R-sig-ME] ordinal regression with MCMCglmm

[David Duffy]

On Wed, 14 Apr 2010, ONKELINX, Thierry wrote:

> Dear Jarrod,
>
> I'm working on a similar problem. Does it makes sense to calculate that
> for the fixed effects only? Something like this:
> pnorm(-Xb),
> pnorm(cp[1] - Xb) - pnorm(Xb)
> pnorm(cp[2] - Xb) - pnorm(cp[1] - Xb)
> 1 - pnorm(cp[2] - Xb)

What Jarrod says, elsethread. But it is interesting/confusing to think 
about what the random effects do to the effective spacing of the 
thresholds eg two groups, at least two thresholds, different 
variances ;)

another 2c, David.

-- 
| 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