GV900 Political Explanation, 2017/2018
30 October, 2018
Homework assignment 2
Due Week 7 (13 November)
Write an R code file named (gv900-HW2.R) to complete the following tasks. The easiest
way to write an R code file is to start with an existing file: Duplicate an existing R code file
that you have (e.g., gv900-week4-JointExercise.R) and modify the file contents accordingly.
Rules
• Submit two files, and two files only. That is, submit (1) the coversheet (ESSAY COVERSHEET
2018-2019.docx, available on Moodle) and (2) your R code file (gv900-HW2.R). Don’t
submit your graph or other outputs. You’ll earn 5 points if you do all of these correctly.
• Make sure that you delete your name from your R code file. You’ll earn 5 points if you
do this correctly.
• Execute everything before you submit (e.g., CTRL + A & CTRL + Return on a Windows
PC; Command + A & Command + Enter on a Mac machine), and make sure your file
runs without an error. I will execute your file to check if you did it. You’ll earn 5 points
if your R file runs without an error.
• Your file must have a proper header. You’ll earn 5 points if you do this correctly.
• Add comments and annotations to everything you do. Try to make your code file look
like my code file. If your code file doesn’t have a proper annotation, you’ll lose 5 points.
Don’t copy and paste all the questions into your R code file, but do show me the question
number for each question. You’ll earn 5 points if you do this correctly.
Tasks (5 points each × 15 = 75 points)
1. Load the “world” dataset (world.csv), and store it as an object named world.data.
2. The data set contains a dummy variable (i.e., a nominal variable with two categories)
named oecd that classifies countries into two groups, OECD member countries and nonmember
countries. One way to describe and summarize the information contained in a
nominal variable is to describe the distribution numerically. As we learned during the
past weeks, we describe the distribution of a nominal variable numerically by creating a
frequency table. Create a frequency table of this variable and store it into a data frame
object ft.oecd. The table has to have three columns: values (initially called “Var1”),
frequency (called “Freq”), and percentage (should be called “Percentage”). Change the
column name of the first column to “OECD Member?”.
3. According to the frequency table you created above, (A) how many countries in the data
set are OECD members? (B) How many countries in the data set are not? (C) What
percentage of countries are OECD members? (D) What percentage of countries are nonmembers?
Give me four answers (four numbers) as a comment. Note: for this task, you
don’t need an R command. Just read the table and tell me the numbers. Don’t forget to
comment them out.
4. Another way to describe and summarize a nominal variable is to draw a frequency distribution
graph. For nominal variables, we draw a bar chart. Using the functions available
in the ggplot2 package (e.g., geom bar), draw a bar chart of the dummy variable that
measures OECD membership.
• Hint 1: Don’t forget to load the package using the library function. It’s usually a
good idea to do so at the beginning of your R code file.
• Hint 2: Don’t forget to change the axis labels using the xlab and ylab options. The
appropriate label for the X axis would be “OECD membership”, whereas the label
for the Y axis could be “Number of countries”.
5. List three countries that are coded as OECD member states. List three countries that
are non-democratic according to the democracy dummy variable. Note: Again, you don’t
need a command for this one; I only need six country names.
6. The data set contains a numerical variable (interval-level variable) named gdp 10 thou
that records a country’s per capita GDP in 10,000 US dollars. Note that this variable
measures per capita GDP in 10,000 dollars, not in dollars. This means that, when this
variable takes a value of 4, for example, then that country’s per cpaita GDP is 40,000
dollars, not 4 dollars. Describe this variable numerically by calculating the following
statistics:
• Range (minimum and maximum), median, mean, 1st and 3rd quartile values (Hint:
this can be done at once with one command)
• Standard deviation (Hint: you need to take care of missing values using the na.rm
option)
Note: You need to provide R commands, not just numerical answers for this one.
7. It appears that the mean and the median of this per capita GDP variable are far apart:
the mean is 6,018 dollars whereas the median is 1,897 dollars. Given that the mean is
much higher than the median, the distribution of this variable is very skewed (i.e., not
symmetric). In which way does the skew go? Answer this question by choosing between
two options: (A) negatively skewed (skewed to the left) or (B) positively skewed (skewed
to the right). Note: Give me your answer in words, not in R commands.
8. Describe this per capita GDP variable graphically by drawing a histogram.
• Hint: Don’t forget to change the axis labels using the xlab and ylab options. The
appropriate label for the X axis would be “Per capita GDP (in 10,000 US dollars)”,
whereas the label for the Y axis could be “Number of countries”.
9. There are two countries in the data set whose per capita GDP is greater than 40,000 US
dollars. Identify these two countries. For this task, I need an R command that gives us
the name of the two countries. Your command will probably generate the symbols
(14 of them), along with the name of the two countries, but that’s fine.
10. We have calculated the sample mean of this per capita GDP variable in task 6. We
have also calculated its standard deviation. We also know from task 6 that there are 14
observations (countries) where this variable is missing, so we have 191 (total number of
countries in the data set) −14 = 177 observations (i.e., n = 177). Therefore, we have all
the building blocks to calculate the standard error of the mean. Calculate the standard
error (the answer should be 0.07091015). Note that I need R commands, not just the
numerical answer.
11. Using the calculated standard error and the mean value, construct the 95 % confidence
interval of the sample mean of gdp 10 thou. For this, I need both R commands and the
numerical answer.
12. Draw histograms of per capita GDP variable, one for democracies and the other for nondemocracies.
• Hint 1: Use the democ regime variable to classify the countries into democracies and
non-democracies.
• Hint 2: Use the facet wrap option. For this task, you may actually have three
histograms (No, Yes, and NA), and that’s OK. We will correct it below.
13. We find (I mean, I find) a few things about this graph unsatisfactory. First, it is a little
bit aesthetically unpleasing that we have a blank graph on the far right. This happens
because there are missing values. Second, the labels “No” and “Yes” are not intuitive at
all (readers can’t know what “Yes” and “No” mean simply by looking at the graph). So
let’s now fix these two things. Create a new data frame named dem.gdp that excludes
those rows where the democ regime variable is missing. Use the is.na function for this.
Then, create a new variable dem.dum within this new data set, which has two nominal
values, “Democracy” and “Autocracy”, instead of “Yes” and “No”. Then, recreate the
histograms you drew in 12. It should look like the following:
Autocracy Democracy
Per capita GDP (in 10,000 US dollars)
Number of countries
14. The graph above appears to suggest that democracies tend to have higher per capita
GDP. Let’s document this relationship by calculating the mean value of per capita GDP
for each group. In doing so, report the 95 % confidence intervals as well. For task 14,
calculate the mean of per capita GDP for democracies, along with the 95 % confidence
interval. Please provide both the commands as well as the results (numbers).
15. Similarly, calculate the mean of per capita GDP for autocracies (non-democracies), along
with the 95 % confidence interval.
End of file