Quantitative Research Exercises, Lesson 5.

1. To refresh your mind, check that gender is a dichotomous variable. If it isn’t, do something!

#load dataset sb12-qr1.sav
library(foreign)
db = file.choose()
dataset = read.spss(db, to.data.frame = TRUE)

/Users/qumino2/Downloads/17_spring_courses/Quantitative_course_做Udacity/exercise_for_Heidi/exercise_data.sav: Unrecognized record type 7, subtype 18 encountered in system file/Users/qumino2/Downloads/17_spring_courses/Quantitative_course_做Udacity/exercise_for_Heidi/exercise_data.sav: Unrecognized record type 7, subtype 21 encountered in system file

table(dataset$Gender)

   0    1 
1382  715 

Based on the result, I think ‘gender’ is a dichotomous variable.

---

2. Assess with graphs and tests if the following variables are normally distributed.

1. Year_birth
2. credits_total
3. q40

#Histogram for Year_birth
ggplot(dataset, aes(Year_birth)) +
  geom_histogram(breaks = seq(1950, 1995, by = 1),
                 col = "red",
                 fill = "green",
                 alpha = .2) +
  labs(title = "Histogram for Year_birth") +
  xlim(c(1949, 1996)) 

  
#qqplot for Year_birth
qqnorm(dataset$Year_birth); qqline(dataset$Year_birth, col =2)

shapiro.test(dataset$Year_birth)

    Shapiro-Wilk normality test

data:  dataset$Year_birth
W = 0.73774, p-value < 2.2e-16

Based on the result, not normally distributed.

#Histogram for credits_total
ggplot(dataset, aes(credits_total)) +
  geom_histogram(binwidth = 10,
                 col = "red",
                 fill = "green",
                 alpha = .2) +
  labs(title = "Histogram for credits_total")

#qqplot for credits_total
qqnorm(dataset$credits_total); qqline(dataset$credits_total, col =2)

shapiro.test(dataset$credits_total)

    Shapiro-Wilk normality test

data:  dataset$credits_total
W = 0.93642, p-value < 2.2e-16

Based on the result, not normally distributed.

#Histogram for q40
ggplot(dataset, aes(q40)) +
  geom_histogram(binwidth = 1,
                 col = "red",
                 fill = "green",
                 alpha = .2) +
  labs(title = "Histogram for q40") 

NA

#qqplot for q40
qqnorm(dataset$q40); qqline(dataset$q40, col =2)

shapiro.test(dataset$q40)

    Shapiro-Wilk normality test

data:  dataset$q40
W = 0.74137, p-value < 2.2e-16

Based on the result, not normally distributed.

---

3. Make a crosstabulation of the following variables. Perform also Chi Square test or Fisher’s Exact test. What can you say about the independency?

1. gender and q12_2
2. gender and q15_7
3. credits_autumn and q12_1. Combine categories in credits_autumn so that you’ll end up with 3-4 different categories.
4. q40 and q28. At first, categorize variable q40.

#a. gender and q12_2
dataset$Gender[dataset$Gender == 0] <- 'Female'
dataset$Gender[dataset$Gender == 1] <- 'Male'
a.tab <- table(dataset$Gender, dataset$q12_2)
a.tab

        
         Strongly disagree Moderately disagree
  Female                15                  25
  Male                  11                  23
        
         Slightly disagree Undecided Slightly agree
  Female                78       232            552
  Male                  58       110            291
        
         Moderately agree Strongly agree
  Female              299            170
  Male                149             69

chisq.test(a.tab)

    Pearson's Chi-squared test

data:  a.tab
X-squared = 12.953, df = 6, p-value = 0.04379

Based on the result, there is a significant difference between male and female’s answers in q12_2

#b. gender and q15_7
b.tab <- table(dataset$Gender, dataset$q15_7)
b.tab

        
         Strongly disagree Moderately disagree
  Female                64                 137
  Male                  20                  72
        
         Slightly disagree Undecided Slightly agree
  Female               421       369            292
  Male                 174       196            180
        
         Moderately agree Strongly agree
  Female               74             22
  Male                 56             11

chisq.test(b.tab)

    Pearson's Chi-squared test

data:  b.tab
X-squared = 18.411, df = 6, p-value = 0.005283

Based on the result, there is a significant difference between male and female’s answers in q15_7

#c. credits_autumn and q12_1. Combine categories in credits_autumn so that you’ll end up
#with 3-4 different categories.
require(car)
dataset$credits_autumn <- recode(dataset$credits_autumn, "c('none', 'Less than 10', '10-14', '15-19') = 'Less than 19'; c('20-24', '25-29', '30-34', '35-39') = '20-39'; c('40-44', '45 or more') = '40 or more'")
table(dataset$credits_autumn)

       20-39   40 or more Less than 19 
        1289          302          508 

c.tab <- table(dataset$credits_autumn, dataset$q12_1)
c.tab

              
               Strongly disagree Moderately disagree
  20-39                      643                 211
  40 or more                 168                  37
  Less than 19               183                  67
              
               Slightly disagree Undecided
  20-39                      176        93
  40 or more                  41        18
  Less than 19                75        51
              
               Slightly agree Moderately agree
  20-39                   127               17
  40 or more               25                4
  Less than 19             84               22
              
               Strongly agree
  20-39                    11
  40 or more                4
  Less than 19             16

chisq.test(b.tab)

    Pearson's Chi-squared test

data:  b.tab
X-squared = 18.411, df = 6, p-value = 0.005283

Based on the result, there is a significant difference

#d. q40 and q28. At first, categorize variable q40.
dataset$q40[dataset_test$q40 <= 10] <- '0-10' 
dataset$q40[dataset_test$q40 > 10 & dataset_test$q40 <= 20] <- '11-20'
dataset$q40[dataset_test$q40 > 20 & dataset_test$q40 <= 40] <- '21-40'
d.tab <- table(dataset$q40, dataset$q28)
d.tab

       
        Yes, constantly Yes, often Sometimes Not often
  0-10               33        160       652       664
  11-20               1         10        45        41
  21-40               0          3         4         5
       
        Not at all
  0-10         392
  11-20         34
  21-40          3

chisq.test(d.tab)

Chi-squared approximation may be incorrect

    Pearson's Chi-squared test

data:  d.tab
X-squared = 5.8199, df = 8, p-value = 0.6674

Based on the result, there is no significant difference