14.09.2015 Views

Matvec Users’ Guide

Matvec Users' Guide

Matvec Users' Guide

SHOW MORE
SHOW LESS
  • No tags were found...

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

70 CHAPTER 11. GENERALIZED LINEAR MIXED MODEL ANALYSES<br />

% P o s i t i v e<br />

1 0 0<br />

9 0<br />

8 0<br />

7 0<br />

6 0<br />

5 0<br />

4 0<br />

3 0<br />

2 0<br />

1 0<br />

0<br />

0 1 2 3 4 5 6<br />

L o g 2 ( D o s e )<br />

F o r m u l a t i o n A B<br />

Figure 11.1: Dose response curves for two different formulations of am antibiotic.<br />

11.1.1 Model Specification<br />

The specification of a generalized linear model consists of descriptors for the linear predictor and the link<br />

function. The linear predictor for our example is:<br />

• µ: intercept<br />

η ij = µ + F i + b x j + b i x j<br />

• F i : main effect of formulation i at x = log 2 (Dose) = 0<br />

• b: linear effect of log 2 (Dose)<br />

• b i : linear interaction between formulation and log 2 (Dose).<br />

In <strong>Matvec</strong> this linear predictor is denoted by:<br />

M=Model(); // Create a Model object.<br />

M.equation("perc pos=intercept Form log2dose Form*log2dose");<br />

// Linear predictor η ij = µ + F i + b x j + b i x j<br />

M.covariate("log2dose");<br />

M.weight("n");<br />

which is the same as a linear model.<br />

Next the logit link function is specified:<br />

M.link("logit",0)<br />

// Logit link, scale parameter is not estimated.<br />

M.fitdata(D);<br />

// Use Data object D<br />

M.link(‘‘fn’’,nvc) specifies both the link function, ‘‘fn’’, and the number of over-dispersion parameters,<br />

nvc, to estimate. Table 11.2 contains a number link functions that are currently available.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!