[R-sig-ME] specifying model with multiple interactions

[James Croft]

I am having trouble specifying a mixed effects model and would  
appreciate some guidance.

Background
We are interested in how cricket batsmen respond to balls delivered  
from bowling machines. We filmed batsmen facing deliveries at  
different speeds from a bower and machine. The data is very expensive  
because each video has to be digitized by hand to identify various  
parameters of interest. Hence we only have data on 13 subjects (and it  
has missing values).

One objective is to predict the timing of bat lift (BatUp) by delivery  
speed (Speed) for a given condition (Delivery) and to assess the  
variability across subjects (Subject).

Delivery  : Factor w/ 2 levels "Bowler","Machine"
Subject   : Factor w/ 13 levels
Speed     : num
BatUp     : num

So I believe I want to predict a fixed effect for speed and random  
slope and intercept for each level of a factor created by  
Subject:Delivery

I'm getting a bit confused with the combinations of random effect.
I have read through a few of the chapters  of Doug Bates. (2010).  
lme4: Mixed-effects modeling with R.

This is as far as I have got...

BatUp ~ Speed + (1|Delivery) + (1|Subject) + (1|Subject:Delivery) +  
(0+Speed|Subject:Delivery)

Thanks in advance for any help.

Regards,
James









Regards,
James


________________________________________
James Croft, PhD
Research Fellow
Sport Performance Research Institute New Zealand
School of Sport and Recreation
AUT University
Private Bag 92006
Auckland 1142
New Zealand

Phone 64-9-921 9999 ext 7685
Fax 64-9 921 9960
Email james.croft at aut.ac.nz
http://www.isrrnz.ac.nz/