> # 6.

> data<-read.csv(file="sorethroat.csv")

>

> data$T.fac<-factor(data$T,levels=c("0","1"))

> data$Y.fac<-factor(data$Y,levels=c("0","1"))

>

> model<-glm(data$Y.fac~data$T.fac+data$D,family="binomial")

> summary(model,cor=T)

Call:

glm(formula = data$Y.fac ~ data$T.fac + data$D, family = "binomial")

Deviance Residuals:

Min 1Q Median 3Q Max

-2.3802 -0.5358 0.3047 0.7308 1.7821

Coefficients:

Estimate Std. Error z value Pr(>|z|)

(Intercept) -1.41734 1.09457 -1.295 0.19536

data$T.fac1 -1.65895 0.92285 -1.798 0.07224 .

data$D 0.06868 0.02641 2.600 0.00931 **

---

Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

(Dispersion parameter for binomial family taken to be 1)

Null deviance: 46.180 on 34 degrees of freedom

Residual deviance: 30.138 on 32 degrees of freedom

AIC: 36.138
Number of Fisher Scoring iterations: 5

Correlation of Coefficients:

(Intercept) data$T.fac1

data$T.fac1 -0.25

data$D -0.83 -0.16

>

> model$null.deviance - model$deviance

[1] 16.04187

>

> step(model,scope=list(lower=formula(model), upper=~.^2),

+ trace=T,scale=1,direction="both")

Start: AIC=36.14

data$Y.fac ~ data$T.fac + data$D

Df Deviance AIC

<none> 30.138 36.138

+ data$T.fac:data$D 1 28.321 36.321

Call: glm(formula = data$Y.fac ~ data$T.fac + data$D, family = "binomial")

Coefficients:

(Intercept) data$T.fac1 data$D

-1.41734 -1.65895 0.06868

Degrees of Freedom: 34 Total (i.e. Null); 32 Residual

Null Deviance: 46.18

Residual Deviance: 30.14 AIC: 36.14

>

>

> # 6.12

> yes<-c(512,89,353,17,120,202,138,131,53,94,22,24)

> no<-c(313,19,207,8,205,391,279,244,138,299,351,317)

> table.6.7<-cbind(expand.grid(gender=c("male","female"),

+ dept=c("A","B","C","D","E","F")),

+ prop=yes/(yes+no))

> table.6.7

gender dept prop

1 male A 0.62060606

2 female A 0.82407407

3 male B 0.63035714

4 female B 0.68000000

5 male C 0.36923077

6 female C 0.34064081

7 male D 0.33093525

8 female D 0.34933333

9 male E 0.27748691

10 female E 0.23918575

11 male F 0.05898123

12 female F 0.07038123

> res.gradadmit<-glm(prop~dept, family=binomial, data=table.6.7, weights=yes+no)

> summary(res.gradadmit)

Call:

glm(formula = prop ~ dept, family = binomial, data = table.6.7,

weights = yes + no)

Deviance Residuals:

Min 1Q Median 3Q Max

-1.4064 -0.4550 0.1456 0.5471 4.1323

Coefficients:

Estimate Std. Error z value Pr(>|z|)

(Intercept) 0.59346 0.06838 8.679 <2e-16 ***

deptB -0.05059 0.10968 -0.461 0.645

deptC -1.20915 0.09726 -12.432 <2e-16 ***

deptD -1.25833 0.10152 -12.395 <2e-16 ***

deptE -1.68296 0.11733 -14.343 <2e-16 ***

deptF -3.26911 0.16707 -19.567 <2e-16 ***

---

Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

(Dispersion parameter for binomial family taken to be 1)

Null deviance: 877.056 on 11 degrees of freedom

Residual deviance: 21.736 on 6 degrees of freedom

AIC: 102.68

Number of Fisher Scoring iterations: 4

> res.gradadmit$deviance

[1] 21.73551

> sum(resid(res.gradadmit, type="pearson")^2)

[1] 19.93841

> resid(res.gradadmit, type="pearson")/sqrt(1-lm.influence(res.gradadmit)$hat)

 1 2 3 4 5 6 7 8 9 10

-4.1530728 4.1530728 -0.5037077 0.5037077 0.8680662 -0.8680662 -0.5458732 0.5458732

1.0005342 -1.0005342

11 12

-0.6197526 0.6197526