The data set Caesar in vcdExtra gives a 3 2 frequency table classifying 251 women who gave birth by Caesarian section by Infection (three levels: none, Type 1, Type2) and Risk, whether Antibiotics were used, and whether the Caesarian section was Planned or not. Infection is a natural response variable. In this exercise, consider only the binary outcome of infection vs. no infection.
library(effects)
## Loading required package: carData
## lattice theme set by effectsTheme()
## See ?effectsTheme for details.
library(vcdExtra)
## Loading required package: vcd
## Loading required package: grid
## Loading required package: gnm
##
## Attaching package: 'vcdExtra'
## The following object is masked from 'package:carData':
##
## Burt
data("Caesar", package = "vcdExtra")
Caesar.df <- as.data.frame(Caesar)
Caesar.df$Infect <- as.numeric(Caesar.df$Infection %in% c('Type 1', 'Type 2'))
str(Caesar.df)
## 'data.frame': 24 obs. of 6 variables:
## $ Infection : Factor w/ 3 levels "Type 1","Type 2",..: 1 2 3 1 2 3 1 2 3 1 ...
## $ Risk : Factor w/ 2 levels "Yes","No": 1 1 1 2 2 2 1 1 1 2 ...
## $ Antibiotics: Factor w/ 2 levels "Yes","No": 1 1 1 1 1 1 2 2 2 2 ...
## $ Planned : Factor w/ 2 levels "Yes","No": 1 1 1 1 1 1 1 1 1 1 ...
## $ Freq : num 0 1 17 0 1 1 11 17 30 4 ...
## $ Infect : num 1 1 0 1 1 0 1 1 0 1 ...
Caesar.df$Risk <- factor(Caesar.df$Risk, levels(Caesar.df$Risk)[c(2,1)])
Caesar.df$Antibiotics <- factor(Caesar.df$Antibiotics, levels(Caesar.df$Antibiotics)[c(2,1)])
Caesar.df$Planned <- factor(Caesar.df$Planned, levels(Caesar.df$Planned)[c(2,1)])
str(Caesar.df)
## 'data.frame': 24 obs. of 6 variables:
## $ Infection : Factor w/ 3 levels "Type 1","Type 2",..: 1 2 3 1 2 3 1 2 3 1 ...
## $ Risk : Factor w/ 2 levels "No","Yes": 2 2 2 1 1 1 2 2 2 1 ...
## $ Antibiotics: Factor w/ 2 levels "No","Yes": 2 2 2 2 2 2 1 1 1 1 ...
## $ Planned : Factor w/ 2 levels "No","Yes": 2 2 2 2 2 2 2 2 2 2 ...
## $ Freq : num 0 1 17 0 1 1 11 17 30 4 ...
## $ Infect : num 1 1 0 1 1 0 1 1 0 1 ...
Caesar.logistic <- glm(Infect ~ Risk + Antibiotics + Planned, data = Caesar.df, family = binomial, weights=Freq)
summary(Caesar.logistic)
##
## Call:
## glm(formula = Infect ~ Risk + Antibiotics + Planned, family = binomial,
## data = Caesar.df, weights = Freq)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -6.7471 -0.4426 0.0000 3.2338 5.4201
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.7935 0.4785 -1.658 0.0972 .
## RiskYes 1.8270 0.4364 4.186 2.84e-05 ***
## AntibioticsYes -3.0011 0.4593 -6.535 6.37e-11 ***
## PlannedYes -0.9064 0.4084 -2.219 0.0265 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 300.85 on 16 degrees of freedom
## Residual deviance: 236.36 on 13 degrees of freedom
## AIC: 244.36
##
## Number of Fisher Scoring iterations: 6
effect <- exp(coef(Caesar.logistic)) - 1
effect.perc <- paste(round(100*effect, 2), "%", sep="")
effect.perc.abs <- paste(round(100*abs(effect), 2), "%", sep="")
effect.perc
## [1] "-54.77%" "521.52%" "-95.03%" "-59.6%"
when an antibiotic is present, the risk of infection decrease.the reduction goes in -95%
effect <- exp(coef(Caesar.logistic)) - 1
effect.perc <- paste(round(100*effect, 2), "%", sep="")
effect.perc.abs <- paste(round(100*abs(effect), 2), "%", sep="")
effect.perc
## [1] "-54.77%" "521.52%" "-95.03%" "-59.6%"
plot(Caesar.logistic)
Caesar.logistic.inter <- update(Caesar.logistic, . ~ . + Risk:Antibiotics)
plot(allEffects(Caesar.logistic.inter))