Data

Reading Data

Original data came from intenet version of Sejong silok, summarized by Oh, Ki-Soo.

sejong_poll <- read.table("../data/sejong_poll.txt", header = TRUE, stringsAsFactors = FALSE)
str(sejong_poll)
## 'data.frame':    44 obs. of  4 variables:
##  $ counts: int  21 194 259 393 443 117 1123 71 29 5 ...
##  $ vote  : chr  "yes" "no" "yes" "no" ...
##  $ class : chr  "high" "high" "third.current" "third.current" ...
##  $ region: chr  "Seoul" "Seoul" "Seoul" "Seoul" ...
# pander(sejong_poll)
kable(sejong_poll[4:1])
region class vote counts
Seoul high yes 21
Seoul high no 194
Seoul third.current yes 259
Seoul third.current no 393
Seoul third.ex yes 443
Seoul third.ex no 117
yuhu ordinary yes 1123
yuhu ordinary no 71
gyunggi chief yes 29
gyunggi chief no 5
gyunggi ordinary yes 17076
gyunggi ordinary no 236
pyungan high no 1
pyungan chief yes 6
pyungan chief no 35
pyungan ordinary yes 1326
pyungan ordinary no 28474
hwanghae chief yes 17
hwanghae chief no 17
hwanghae ordinary yes 4454
hwanghae ordinary no 15601
chungcheong high no 2
chungcheong chief yes 35
chungcheong chief no 26
chungcheong ordinary yes 6982
chungcheong ordinary no 14013
kangwon chief yes 5
kangwon chief no 10
kangwon ordinary yes 939
kangwon ordinary no 6888
hamgil high no 1
hamgil chief yes 3
hamgil chief no 14
hamgil ordinary yes 75
hamgil ordinary no 7387
gyungsang chief yes 55
gyungsang chief no 16
gyungsang ordinary yes 36262
gyungsang ordinary no 377
jeolla high no 2
jeolla chief yes 42
jeolla chief no 12
jeolla ordinary yes 29505
jeolla ordinary no 257

Factor conversion

We need vote, class, region as factors. If you leave them as chr, it will be coerced to factor when you tabulate it according to alphabetical order, which is not what you want. So, use factor() to convert them. First, make a working copy vesion of sejong_poll

sejong_poll_2 <- sejong_poll
sejong_poll_2$vote <- factor(sejong_poll_2$vote, levels = c("yes","no"), labels = c("Yes", "No"))

You can check that labels = is not necessary if same as levels. Continue with class and region_

class_levels <- c("high","third.current", "third.ex", "chief", "ordinary")
class_labels <- c("High","3rd_current", "3rd_former", "Chief", "Commons")
sejong_poll_2$class <- factor(sejong_poll_2$class, levels = class_levels, labels = class_labels)
region_levels <- c("Seoul","yuhu", "gyunggi", "pyungan", "hwanghae", "chungcheong", "kangwon", "hamgil", "gyungsang", "jeolla")
# region_labels <- c("Seoul","Yuhu", "Gyunggi", "Pyungan", "Hwanghae", "Chungcheong", "Kangwon", "Hamgil", "Gyungsang", "Jeolla")
region_labels <- c("SL","YH", "GG", "PA", "HH", "CC", "KW", "HG", "GS", "JL")
sejong_poll_2$region <- factor(sejong_poll_2$region, levels = region_levels, labels = region_labels)
str(sejong_poll_2)
## 'data.frame':    44 obs. of  4 variables:
##  $ counts: int  21 194 259 393 443 117 1123 71 29 5 ...
##  $ vote  : Factor w/ 2 levels "Yes","No": 1 2 1 2 1 2 1 2 1 2 ...
##  $ class : Factor w/ 5 levels "High","3rd_current",..: 1 1 2 2 3 3 5 5 4 4 ...
##  $ region: Factor w/ 10 levels "SL","YH","GG",..: 1 1 1 1 1 1 2 2 3 3 ...
kable(sejong_poll_2[4:1])
region class vote counts
SL High Yes 21
SL High No 194
SL 3rd_current Yes 259
SL 3rd_current No 393
SL 3rd_former Yes 443
SL 3rd_former No 117
YH Commons Yes 1123
YH Commons No 71
GG Chief Yes 29
GG Chief No 5
GG Commons Yes 17076
GG Commons No 236
PA High No 1
PA Chief Yes 6
PA Chief No 35
PA Commons Yes 1326
PA Commons No 28474
HH Chief Yes 17
HH Chief No 17
HH Commons Yes 4454
HH Commons No 15601
CC High No 2
CC Chief Yes 35
CC Chief No 26
CC Commons Yes 6982
CC Commons No 14013
KW Chief Yes 5
KW Chief No 10
KW Commons Yes 939
KW Commons No 6888
HG High No 1
HG Chief Yes 3
HG Chief No 14
HG Commons Yes 75
HG Commons No 7387
GS Chief Yes 55
GS Chief No 16
GS Commons Yes 36262
GS Commons No 377
JL High No 2
JL Chief Yes 42
JL Chief No 12
JL Commons Yes 29505
JL Commons No 257

Array

We can set up the data as an array

sejong_poll_array <- xtabs(counts ~ vote + class + region, 
                           data = sejong_poll_2)
str(sejong_poll_array)
##  'xtabs' int [1:2, 1:5, 1:10] 21 194 259 393 443 117 0 0 0 0 ...
##  - attr(*, "dimnames")=List of 3
##   ..$ vote  : chr [1:2] "Yes" "No"
##   ..$ class : chr [1:5] "High" "3rd_current" "3rd_former" "Chief" ...
##   ..$ region: chr [1:10] "SL" "YH" "GG" "PA" ...
##  - attr(*, "call")= language xtabs(formula = counts ~ vote + class + region, data = sejong_poll_2)
sejong_poll_array
## , , region = SL
## 
##      class
## vote   High 3rd_current 3rd_former Chief Commons
##   Yes    21         259        443     0       0
##   No    194         393        117     0       0
## 
## , , region = YH
## 
##      class
## vote   High 3rd_current 3rd_former Chief Commons
##   Yes     0           0          0     0    1123
##   No      0           0          0     0      71
## 
## , , region = GG
## 
##      class
## vote   High 3rd_current 3rd_former Chief Commons
##   Yes     0           0          0    29   17076
##   No      0           0          0     5     236
## 
## , , region = PA
## 
##      class
## vote   High 3rd_current 3rd_former Chief Commons
##   Yes     0           0          0     6    1326
##   No      1           0          0    35   28474
## 
## , , region = HH
## 
##      class
## vote   High 3rd_current 3rd_former Chief Commons
##   Yes     0           0          0    17    4454
##   No      0           0          0    17   15601
## 
## , , region = CC
## 
##      class
## vote   High 3rd_current 3rd_former Chief Commons
##   Yes     0           0          0    35    6982
##   No      2           0          0    26   14013
## 
## , , region = KW
## 
##      class
## vote   High 3rd_current 3rd_former Chief Commons
##   Yes     0           0          0     5     939
##   No      0           0          0    10    6888
## 
## , , region = HG
## 
##      class
## vote   High 3rd_current 3rd_former Chief Commons
##   Yes     0           0          0     3      75
##   No      1           0          0    14    7387
## 
## , , region = GS
## 
##      class
## vote   High 3rd_current 3rd_former Chief Commons
##   Yes     0           0          0    55   36262
##   No      0           0          0    16     377
## 
## , , region = JL
## 
##      class
## vote   High 3rd_current 3rd_former Chief Commons
##   Yes     0           0          0    42   29505
##   No      2           0          0    12     257

Votes

Total

Check the total vote with xtabs()

vote_total <- xtabs(counts ~ vote, 
                    data = sejong_poll_2)
kable(t(as.matrix(vote_total)), 
      caption = "Total")
Total
Yes No
98657 74149 
# format(prop.table(vote_total)*100, digits = 3, nsmall = 1)
kable(t(as.matrix(format(prop.table(vote_total) * 100, 
                         digits = 3, 
                         nsmall = 1))), 
      caption = "Percentage", 
      align = rep("r", 2))
Percentage
Yes No
57.1 42.9 
vote_total.2 <- apply(sejong_poll_array, 1, sum)
# kable(t(as.matrix(vote_total.2)))
kable(t(as.matrix(vote_total.2)), 
      caption = "Total")
Total
Yes No
98657 74149

Vote by class

vote_class <- xtabs(counts ~ vote + class, 
                    data = sejong_poll_2)
kable(vote_class, 
      caption = "By Class")
By Class
High 3rd_current 3rd_former Chief Commons
Yes 21 259 443 192 97742
No 200 393 117 135 73304 
vote_class_a <- apply(sejong_poll_array, 1:2, sum)
kable(vote_class_a, 
      caption = "By Class")
By Class
High 3rd_current 3rd_former Chief Commons
Yes 21 259 443 192 97742
No 200 393 117 135 73304

Commons vs Bureaucrats

We need to analyse Commons separately.

sejong_poll_2$class_2 <- factor(ifelse(sejong_poll_2$class == "Commons", 
                                       "Commons", "Bureaus"), 
                                levels = c("Bureaus", "Commons"))
kable(sejong_poll_2[c(4, 3, 5, 2, 1)])
region class class_2 vote counts
SL High Bureaus Yes 21
SL High Bureaus No 194
SL 3rd_current Bureaus Yes 259
SL 3rd_current Bureaus No 393
SL 3rd_former Bureaus Yes 443
SL 3rd_former Bureaus No 117
YH Commons Commons Yes 1123
YH Commons Commons No 71
GG Chief Bureaus Yes 29
GG Chief Bureaus No 5
GG Commons Commons Yes 17076
GG Commons Commons No 236
PA High Bureaus No 1
PA Chief Bureaus Yes 6
PA Chief Bureaus No 35
PA Commons Commons Yes 1326
PA Commons Commons No 28474
HH Chief Bureaus Yes 17
HH Chief Bureaus No 17
HH Commons Commons Yes 4454
HH Commons Commons No 15601
CC High Bureaus No 2
CC Chief Bureaus Yes 35
CC Chief Bureaus No 26
CC Commons Commons Yes 6982
CC Commons Commons No 14013
KW Chief Bureaus Yes 5
KW Chief Bureaus No 10
KW Commons Commons Yes 939
KW Commons Commons No 6888
HG High Bureaus No 1
HG Chief Bureaus Yes 3
HG Chief Bureaus No 14
HG Commons Commons Yes 75
HG Commons Commons No 7387
GS Chief Bureaus Yes 55
GS Chief Bureaus No 16
GS Commons Commons Yes 36262
GS Commons Commons No 377
JL High Bureaus No 2
JL Chief Bureaus Yes 42
JL Chief Bureaus No 12
JL Commons Commons Yes 29505
JL Commons Commons No 257 
str(sejong_poll_2)
## 'data.frame':    44 obs. of  5 variables:
##  $ counts : int  21 194 259 393 443 117 1123 71 29 5 ...
##  $ vote   : Factor w/ 2 levels "Yes","No": 1 2 1 2 1 2 1 2 1 2 ...
##  $ class  : Factor w/ 5 levels "High","3rd_current",..: 1 1 2 2 3 3 5 5 4 4 ...
##  $ region : Factor w/ 10 levels "SL","YH","GG",..: 1 1 1 1 1 1 2 2 3 3 ...
##  $ class_2: Factor w/ 2 levels "Bureaus","Commons": 1 1 1 1 1 1 2 2 1 1 ...

Compare the votes by class_2, (Bureaucrats vs Commons)

vote_class_2 <- xtabs(counts ~ vote + class_2, 
                      data = sejong_poll_2)
kable(vote_class_2, caption = "By Bureaus and Commons")
By Bureaus and Commons
Bureaus Commons
Yes 915 97742
No 845 73304 
vote_class_2_a <- cbind("Bureaus" = rowSums(vote_class_a[, -5]), "Commons" =  vote_class_a[, 5])
kable(vote_class_2_a, caption = "By Bureaus and Commons")
By Bureaus and Commons
Bureaus Commons
Yes 915 97742
No 845 73304

Add subtotals to the margins,

vote_class_2_am <- addmargins(vote_class_2)
kable(vote_class_2_am)
Bureaus Commons Sum
Yes 915 97742 98657
No 845 73304 74149
Sum 1760 171046 172806

Compute the marginal proportions. Note the use of digits = 3 and nsmall = 1.

kable(format(prop.table(vote_class_2, margin = 2)*100, digits = 3, nsmall = 1), caption = "Bureaus and Commons", align = rep("r", 2))
Bureaus and Commons
Bureaus Commons
Yes 52.0 57.1
No 48.0 42.9

Votes by region with respect to class_2

Count the vote by region class_2 wise.

class_2 <- sejong_poll_2$class_2
vote_region_bureaus <- xtabs(counts ~ vote + region, 
                             data = sejong_poll_2, 
                             class_2 == "Bureaus", 
                             drop = TRUE)
kable(vote_region_bureaus, caption = "Votes(Bureaus)")
Votes(Bureaus)
SL GG PA HH CC KW HG GS JL
Yes 723 29 6 17 35 5 3 55 42
No 704 5 36 17 28 10 15 16 14 
# xtabs(counts ~ vote + region, data = sejong_poll_2[class_2 == "Bureaus", ], drop = TRUE)
vote_region_commons <- xtabs(counts ~ vote + region, data = sejong_poll_2, class_2 == "Commons", drop = TRUE)
kable(vote_region_commons, caption = "Votes(Commons)")
Votes(Commons)
YH GG PA HH CC KW HG GS JL
Yes 1123 17076 1326 4454 6982 939 75 36262 29505
No 71 236 28474 15601 14013 6888 7387 377 257

Seoul has three times more Bureaucrats than other regions, so analyse further.

region <- sejong_poll_2$region
vote_seoul_class <- xtabs(counts ~ vote + class, data = sejong_poll_2, region == "SL", drop = TRUE)
kable(vote_seoul_class, caption = "Seoul")
Seoul
High 3rd_current 3rd_former
Yes 21 259 443
No 194 393 117 
kable(format(prop.table(vote_seoul_class, margin = 2)*100, digits  = 3, nsmall = 1), caption = "SL", align = rep("r", 3))
SL
High 3rd_current 3rd_former
Yes 9.77 39.72 79.11
No 90.23 60.28 20.89

Chungcheong’s case.

vote_chung_class <- xtabs(counts ~ vote + class, data = sejong_poll_2, region == "CC", drop = TRUE)
kable(format(prop.table(vote_chung_class, margin = 2)*100, digits = 3, nsmall = 1), caption = "CC", align = rep("r", 3))
CC
High Chief Commons
Yes 0.0 57.4 33.3
No 100.0 42.6 66.7
  • Save the working directory image.
save.image(file = "sejong_poll_data.RData")