> berk1.df <- read.table("data/berkeley.dat",header=T)
> attach(berk1.df)
> D <- matrix(freq,byrow=T,ncol=4)
> or <- function(x) { ((x[1]+.5)/(x[2]+.5))/((x[3]+.5)/(x[4]+.5)) }
> cbind(1:6,round(apply(D,1,or),2)) # odds-ratios by department
     [,1] [,2]
[1,]    1 0.34
[2,]    2 0.83
[3,]    3 1.13
[4,]    4 0.92
[5,]    5 1.22
[6,]    6 0.83
> dept <- rep(c("A","B","C","D","E","F"),each=2)
> sex <- rep(c("M","F"),6)
> admit <- freq[(0:11)*2+1]
> deny <- freq[(1:12)*2]
> berk2.df <- data.frame(dept,sex,admit,deny)
> berk2.df
   dept sex admit deny
1     A   M   512  331
2     A   F    89   19
3     B   M   353  207
4     B   F    17    8
5     C   M   120  205
6     C   F   202  391
7     D   M   138  279
8     D   F   131  244
9     E   M    53  138
10    E   F    94  299
11    F   M    22  351
12    F   F    24  317


> fit1 <- glm(cbind(admit,deny)~dept+sex,binomial,data=berk2.df)
> summary(fit1)

Call:
glm(formula = cbind(admit, deny) ~ dept + sex, family = binomial, 
    data = berk2.df)

Deviance Residuals: 
 [1]  -1.30458   3.92943  -0.05206   0.25174   1.30003  -0.95797   0.12110
 [8]  -0.12568   1.25470  -0.87356  -0.18836   0.18574

Coefficients:
             Estimate Std. Error z value Pr(>|z|)    
(Intercept)  0.636739   0.098409    6.47 9.78e-11 ***
deptB        0.009772   0.109123    0.09    0.929    
deptC       -1.214525   0.105981  -11.46  < 2e-16 ***
deptD       -1.245105   0.105149  -11.84  < 2e-16 ***
deptE       -1.691493   0.125576  -13.47  < 2e-16 ***
deptF       -3.257133   0.169577  -19.21  < 2e-16 ***
sexM        -0.108203   0.080775   -1.34    0.180    
---
Signif. codes:  0 `***' 0.001 `**' 0.01 `*' 0.05 `.' 0.1 ` ' 1 

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 860.409  on 11  degrees of freedom
Residual deviance:  22.254  on  5  degrees of freedom
AIC: 105.23

Number of Fisher Scoring iterations: 3

> anova(fit1)
Analysis of Deviance Table

Model: binomial, link: logit

Response: cbind(admit, deny)

Terms added sequentially (first to last)


     Df Deviance Resid. Df Resid. Dev
NULL                    11     860.41
dept  5   836.35         6      24.06
sex   1     1.80         5      22.25



> fit2 <- glm(cbind(admit,deny)~dept+sex,binomial,subset(berk2.df,dept!="A"))
> summary(fit2)

Call:
glm(formula = cbind(admit, deny) ~ dept + sex, family = binomial, 
    data = subset(berk2.df, dept != "A"))

Deviance Residuals: 
 [1]  -0.1191   0.5680   0.5239  -0.3914  -0.5164   0.5440   0.6868  -0.4892
 [9]  -0.5024   0.5158

Coefficients:
            Estimate Std. Error z value Pr(>|z|)    
(Intercept)  0.51349    0.11936   4.302 1.69e-05 ***
deptC       -1.14008    0.12188  -9.354  < 2e-16 ***
deptD       -1.19456    0.11984  -9.968  < 2e-16 ***
deptE       -1.61308    0.13928 -11.581  < 2e-16 ***
deptF       -3.20527    0.17880 -17.927  < 2e-16 ***
sexM         0.03069    0.08676   0.354    0.724    
---
Signif. codes:  0 `***' 0.001 `**' 0.01 `*' 0.05 `.' 0.1 ` ' 1 

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 539.4581  on 9  degrees of freedom
Residual deviance:   2.5564  on 4  degrees of freedom
AIC: 71.79

Number of Fisher Scoring iterations: 3

> anova(fit2)
Analysis of Deviance Table

Model: binomial, link: logit

Response: cbind(admit, deny)

Terms added sequentially (first to last)


     Df Deviance Resid. Df Resid. Dev
NULL                     9     539.46
dept  4   536.78         5       2.68
sex   1     0.13         4       2.56