[R-sig-ME] Do I need random effects?

[David Duffy]

On Sat, 26 Feb 2011, Murray Jorgensen wrote:

> One or two glitches remain:
>
>> anova(m1, m0)
> Data: Pastes
> Models:
> m1: strength ~ (1 | batch)
> m0: strength ~ (1 | dummy)
>   Df    AIC    BIC  logLik Chisq Chi Df Pr(>Chisq)
> m1  3 308.60 314.88 -151.30
> m0  3 316.26 322.55 -155.13     0      0          1
>
> The df's are not what we want them to be - and I am rather 
> surprised to find a p-value. :-)

Pardon me.  I first ran

m1: strength ~ (1|dummy) + (1 | batch)

which is better handled by  anova(), and then simplified
it without fully checking it.  The value of
the LRT in the second bit was correct, though ;)


-- 
| 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