[R-lang] Re: p-values for mixed effects models with random slopes

[Hugo Quené]

Dear list,

On 2011.01.18 19:50 , Levy, Roger wrote:
> Just some notes on what everyone else has written<see below>:
> I think that there are a few basic options for the general problem of assessing significance of fixed effects in models with rich random-effects structure:
>
> 1) trust the normal approximation to the t- (and more generally the F-) statistic.  My consistent impression is that for models with parameters numbering in the low dozens and observations numbering in the hundreds or thousands, the anti-conservativity is minimal.  The examples from Pinheiro&  Bates 2000 showing anticonservativity have far fewer observations per parameter.

I share this impression. Just to expand a bit:

In his textbook, Hox (2010, p.46), citing Bryk & Raudenbusch (1992), 
mentions the t-test for testing a fixed effect, with df=J-p-1, where 
J=number of higher-level units (e.g. subjects or items, whichever 
the lowest), and p=total number of parameters in model, fixed and 
random combined. Minimum df=1 of course. This is indeed a 
conservative approach, but I find it useful for assessing and 
reporting large effects.

Hox, J.J. (2010). Multilevel Analysis: Techniques and Applications 
(2nd ed.). Mahwah, NJ: Lawrence Erlbaum.

http://www.joophox.net/mlbook2/MLbook.htm
see also Google Books for preview

Best wishes, Hugo

-- 
Dr Hugo Quené | Assoc Prof Phonetics | Linguistics Program | Utrecht 
inst of Linguistics OTS | Universiteit Utrecht | Trans 10 | 3512 JK 
Utrecht | The Netherlands | T +31 30 253 6070 | H.Quene@uu.nl | 
www.hugoquene.nl | uu.academia.edu/HugoQuene | www.hum.uu.nl