Exercise 4.2 (a) Taking support for abortion as the outcome variable, produce fourfold displays showing the association with sex, stratified by status.
library(vcd)
## Loading required package: grid
library(vcdExtra)
## Loading required package: gnm
data("Abortion", package = "vcdExtra")
str(Abortion)
## 'table' num [1:2, 1:2, 1:2] 171 152 138 167 79 148 112 133
## - attr(*, "dimnames")=List of 3
## ..$ Sex : chr [1:2] "Female" "Male"
## ..$ Status : chr [1:2] "Lo" "Hi"
## ..$ Support_Abortion: chr [1:2] "Yes" "No"
#stratified by status
fourfoldplot(aperm(Abortion, c(3,1,2)))
fourfoldplot(aperm(Abortion, c(3,2,1)))
#calculating odds ratio when stratified by status
odds_ratio_by_status<-oddsratio(Abortion, stratum="Status", log=FALSE)
odds_ratio_by_status
## odds ratios for Sex and Status by Support_Abortion
##
## Lo Hi
## 1.3614130 0.6338682
#calculating odds ratio when stratified by sex
odds_ratio_by_sex<-oddsratio(Abortion, stratum="Sex", log=FALSE)
odds_ratio_by_sex
## odds ratios for Sex and Status by Support_Abortion
##
## Female Male
## 1.3614130 0.6338682
Summary From both the stratified graphs (by status and by sex) we can see when the status is “Low” higher proportion of women are supporting abortion. On the contrary, when status is “high” higher proportion of Men are supporting abortion.
Exercise 4.7 (a) Assess the strength of agreement between the raters using Cohen’s ??, both unweighted and weighted.
#creating a matrix with the movie ratings
rating_dataset<- cbind(c(24, 8, 10),c(8,9,13),c(13,10,64))
#lebel the rows and columns
rownames(rating_dataset) = c("Con","Mixed","Pro")
colnames(rating_dataset) = c("Con","Mixed","Pro")
Kappa(rating_dataset)
## value ASE z Pr(>|z|)
## Unweighted 0.3433 0.06042 5.682 1.330e-08
## Weighted 0.4015 0.06428 6.245 4.225e-10
Comment the unweighted value is 0.3433 which shows the stregnth as “fair agreement”. Weighted value is slightly higher 0.4014 which falls into “moderate agreement” range.
Also, p-value is statisitcally significant, this shows that there is considerable agreement between two reviewrs
#lets plot both weighted and unweighted plots and analyze the observations
agreementplot(rating_dataset, main = "Weighted")
agreementplot(rating_dataset, main = "Unweighted")
Comment
By observing the plots we can see the association is higher in “pro” and “con scenarios”, there seems to be relatively less agreement in “con”. This explains the overall kappa values which accounts for all the three scnearios and shows “fair agreement” overall.