TP1.R

##
# TP1
# Simon Perdrisat & Amantin Baruti- 2016
# Replication table 4b p 494 of "ECONOMIC AND POLITICAL EXPLANATIONS OF HUMAN RIGHTS VIOLATIONS" by NEIL J. MITCHELL and JAMES M. McCORMICK
#
# Note: I'm using = instead of <-
# http://blog.revolutionanalytics.com/2008/12/use-equals-or-arrow-for-assignment.html
#
##



dataset =
  read.table(
    "/Users/simon/Documents/school/uni/cours uni/année 2 semestre 1/Méthodes appliquées au domaine international/tp1/Mitchell_et_al_1984_with_edu_index.csv",
    header = TRUE,
    sep = ",",
    quote = "",
    na.strings = "NA",
    dec = ".",
    strip.white = TRUE
  )

# Remove duplicate observation
dataset = dataset[!duplicated(dataset[, c("CCode")]), ]


# Copy the torture_score in a new col
dataset$torture_score_new = dataset$torture_score

# Add a level to the factor torture_score_new so we can t
levels(dataset$torture_score_new) = c(levels(dataset$torture_score_new), "rarely or never")

### Recode and order variables
# Torture score
dataset$torture_score_new[dataset$torture_score == "rarely"] = "rarely or never"
dataset$torture_score_new[dataset$torture_score == "never"] = "rarely or never"

dataset$torture_score_new = droplevels(dataset$torture_score_new)
dataset$torture_score = ordered(dataset$torture_score, levels =c("never", "rarely", "sometimes", "often", "very often"))

dataset$torture_score_new = ordered(dataset$torture_score_new, levels =c("rarely or never", "sometimes", "often", "very often"))



### Text
dataset$colonialExperience = ifelse(dataset$british_colony == "yes",
                                    "Former British Colony",
                                    "Other Colonial Background")

## Use of torture

print("---- Use of torture ----")

useOfTortureByColonialExperience = table(dataset$torture_score_new, dataset$colonialExperience)

# Get the number of observation
addmargins(useOfTortureByColonialExperience,1);

# Contingency table with conditional probability

useOfTortureByColonialExperienceProp = prop.table(useOfTortureByColonialExperience,2)

print(addmargins(useOfTortureByColonialExperienceProp*100,1),3)

## TEST
chisq.test(useOfTortureByColonialExperience)

cramersV(useOfTortureByColonialExperience)

## SANS Asie pacifique

print("---- Use of torture WITOUTH ASIA----")


datasetWithoutAsia = subset(dataset, dataset$continent != "Asia")

useOfTortureByColonialExperienceWithoutAsia = table(datasetWithoutAsia$torture_score_new, datasetWithoutAsia$colonialExperience)

# Get the number of observation
addmargins(useOfTortureByColonialExperienceWithoutAsia,1);

# Contingency table with conditional probability

useOfTortureByColonialExperiencePropWithoutAsia = prop.table(useOfTortureByColonialExperienceWithoutAsia,2)

print(addmargins(useOfTortureByColonialExperiencePropWithoutAsia*100,1),3)

## TEST
chisq.test(useOfTortureByColonialExperienceWithoutAsia)


cramersV(useOfTortureByColonialExperienceWithoutAsia)

### ----
# With education index
### ---


boxplot(dataset$EducationIndex ~ dataset$torture_score,
        xlab="Use of Torture",
        ylab='Education Index',
        main="Use of Torture by Education Index")

# ANOVA TEST

oneway.test(dataset$EducationIndex ~ dataset$torture_score)

## Try to do some kind of regression with those ugly data
dataset$torture_score_int[dataset$torture_score == "never"] = 0
dataset$torture_score_int[dataset$torture_score == "rarely"] = 10
dataset$torture_score_int[dataset$torture_score == "somtimes"] = 55
dataset$torture_score_int[dataset$torture_score == "often"] = 500
dataset$torture_score_int[dataset$torture_score == "very often"] = 1000

lm.out = lm(dataset$EducationIndex ~ dataset$torture_score_int)

oneway.test(dataset$EducationIndex ~ dataset$torture_score_int)

plot(dataset$torture_score_int, dataset$EducationIndex, xlab="Use of Torture", ylab="Education Index")

abline(lm.out, col="red")