Using R and RStudio for Data Management, Statistical Analysis, and Graphics, 2nd Edition

i

i

“K23166” — 2015/1/28 — 9:35 — page 106 — #132

i

i

i

i

i

i

106 CHAPTER 7. REGRESSION GENERALIZATIONS AND MODELING

AIC: 8425

Number of Fisher Scoring iterations: 6

It is always important to check assumptions for models. This is particularly true for Poisson

models, which are quite sensitive to model departures. There is support in the vcd package

for a Pearson’s χ

2

goodness-of-fit test.

> library(vcd)

Loading required package: grid

> poisfit = with(ds, goodfit(e2b, "poisson"))

> summary(poisfit)

Goodness-of-fit test for poisson distribution

X^2 df P(> X^2)

Likelihood Ratio 208 10 3.6e-39

The results indicate that the fit is poor (χ

2

10

= 208, p < 0.0001); the Poisson ...