next up previous
Next: About this document ...

> data <- read.table("sex.dat", header=T)
> attach(data, 1)
> options(contrasts=c("contr.treatment"))
> uv <- sex*birth
> sex <- factor(sex); birth <- factor(birth)
> > indep.fit <- glm(count ~ sex + birth, family=poisson)
> pearson <- summary.lm(indep.fit)$residuals
> hat <- lm.influence(indep.fit)$hat
> adj.res <- pearson/sqrt(1 - hat) # standardized Pearson residuals
> cbind(sex, birth, count, fitted(indep.fit), adj.res)
   sex birth count              adj.res 
 1   1     1    81  42.41145  7.6028389
 2   1     2    68  51.21382  3.0766897
 3   1     3    60  86.42333 -4.1166720
 4   1     4    38  66.95140 -4.8396108
 5   2     1    24  15.96868  2.3282276
 6   2     2    26  19.28294  1.8114763
 7   2     3    29  32.53996 -0.8114805
 8   2     4    14  25.20842 -2.7568090
 9   3     1    18  30.04860 -2.6816389
10   3     2    41  36.28510  0.9762279
11   3     3    74  61.23110  2.2472882
12   3     4    42  47.43521 -1.0263685
13   4     1    36  70.57127 -6.0630727
14   4     2    57  85.21814 -4.6038421
15   4     3   161 143.80562  2.3845449
16   4     4   157 111.40497  6.7845091
> LL.fit <- update(indep.fit, . ~ . + uv)
> summary(LL.fit)

Coefficients:
                 Value Std. Error    t value 
(Intercept)  4.1068414 0.08950939  45.881681
       sex2 -1.6459643 0.13472875 -12.216875
       sex3 -1.7700229 0.16463639 -10.751104
       sex4 -1.7536852 0.23430240  -7.484708
     birth2 -0.4641050 0.11952096  -3.883043
     birth3 -0.7245224 0.16199835  -4.472406
     birth4 -1.8796640 0.24908191  -7.546369
         uv  0.2858355 0.02823566  10.123209

Residual Deviance: 11.53369 on 8 degrees of freedom
Number of Fisher Scoring Iterations: 3 

> u <- c(1,1,1,1,2,2,2,2,4,4,4,4,5,5,5,5)
> v <- c(1,2,4,5,1,2,4,5,1,2,4,5,1,2,4,5)
> uv2 <- u*v
> LL2.fit <- update(indep.fit, . ~ . + uv2)
> summary(LL2.fit)

Coefficients:
                 Value Std. Error    t value 
        uv2  0.1460382 0.01407636  10.374715

Residual Deviance: 8.845195 on 8 degrees of freedom

> row.fit <- update(indep.fit, .~. +u:birth)
> summary(row.fit)

                 Value Std. Error    t value 
(Intercept)  4.9872158 0.14624227  34.102424
       sex2 -0.6577204 0.13124054  -5.011564
       sex3  0.4666360 0.16265012   2.868956
       sex4  1.5019483 0.17951515   8.366694
     birth2 -0.3193886 0.19821146  -1.611353
     birth3 -0.7268794 0.20015708  -3.631545
     birth4 -1.4903163 0.23744302  -6.276522
    birth1u -0.5953263 0.06555007  -9.082009
    birth2u -0.4054336 0.06067887  -6.681628
    birth3u -0.1297489 0.05633659  -2.303102
    birth4u         NA         NA         NA

Residual Deviance: 8.262966 on 6 degrees of freedom

> column.fit <- update(indep.fit, . ~ . + sex:v)
> summary(column.fit)

Coefficients: (1 not defined because of singularities)
                 Value Std. Error    t value 
(Intercept)  4.9151850 0.13907409  35.342204
       sex2 -1.2186815 0.25582686  -4.763696
       sex3 -1.5060354 0.23760340  -6.338442
       sex4 -1.3955940 0.20728940  -6.732587
     birth2  0.5458980 0.11723030   4.656629
     birth3  1.5926216 0.14786815  10.770553
     birth4  1.5101843 0.16419368   9.197579
      sex1v -0.5845440 0.05929738  -9.857840
      sex2v -0.4955406 0.07989777  -6.202182
      sex3v -0.2031464 0.06537858  -3.107231
      sex4v         NA         NA         NA

Residual Deviance: 7.586124 on 6 degrees of freedom





Alan Agresti 2001-12-28