The data used in this exercise were originally reported by Douglas et al (2004) and the data was reproduced by West et al (2012). The main goal of their experiment was to examine nucleotide activation (guanine nucleotide bonding) in seven different brain nuclei (i.e . brain regions) among five adults male rats The basal nucleotide activation, measured after treatment with saline solution, was compared to activation in the same region after treatment with the drug carbachol. Activation was measured as the mean optical density produced by autoradiography. We want to compare activation in a subset of three of the original seven brain regions studied by the authors: the bed nucleus of the stria terminals (BST), the lateral septum (LS) , and the diagonal band of Broca (VDB). The original data layout for this study is shown below. Use an appropriate statistical model toexamine the nucleotide activation.

Data format

Animal: Unique identifier for each rat

Treatment:

(1 = Basal, 2 = Carbachol)

Region:

Brain nucleus (1 = BST, 2 = LS, 3 = VDB)

Dependent variable: activate

Nucleotide activation

1. Change this data set to long format

Wide format

Animal BST LS VDB TreatmentR111097371.71302.02449.7CarbR111397492.58355.74459.58CarbR100797664.72587.1726.96CarbR100997515.29437.56604.29CarbR110597589.25493.93621.07Carb

R111097 366.19 199.31 187.11 Basal

R111397 375.58204.85179.38 BasalR100797 458.16245.04237.42 Basal

R100997 479.81 261.19 195.51 Basal

R110597 462.79278.33262.05 Basal

Your long format has to look like this

The long format has to look like the following

animal treatment region activate

R111097 1 1 371.71

R111097 1 2 302.02

R111097 1 3 449.70

Etc

1. Use an appropriate statistical model toexamine the nucleotide activation, since I did not go over mixed model, here is the code for the mixed model

title "Mixed model";proc mixed data = ratbrain covtest; class animal region; model activate = region treat region*treat / residual solution outpred=resids; random int treat / subject = animal type = un solution v vcorr;run;

or you can use R code

library(lme4)

model.fit <- lmer(activate ~ region treat region*treat + (1|animal),

data = rat.brain, REML = T)

summary(model.fit)