HW the 2nd
Meryem Karayunusoglu
2024-01-29
R Markdown
Setting up your environment
We always need to do this step first and include all the packages we need
# List of packages
packages <- c("tidyverse", "fst", "modelsummary", "viridis") # add any you need here
# Install packages if they aren't installed already
new_packages <- packages[!(packages %in% installed.packages()[,"Package"])]
if(length(new_packages)) install.packages(new_packages)
# Load the packages
lapply(packages, library, character.only = TRUE)## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr 1.1.4 ✔ readr 2.1.5
## ✔ forcats 1.0.0 ✔ stringr 1.5.1
## ✔ ggplot2 3.4.4 ✔ tibble 3.2.1
## ✔ lubridate 1.9.3 ✔ tidyr 1.3.1
## ✔ purrr 1.0.2
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors
## Loading required package: viridisLite## [[1]]
## [1] "lubridate" "forcats" "stringr" "dplyr" "purrr" "readr"
## [7] "tidyr" "tibble" "ggplot2" "tidyverse" "stats" "graphics"
## [13] "grDevices" "utils" "datasets" "methods" "base"
##
## [[2]]
## [1] "fst" "lubridate" "forcats" "stringr" "dplyr" "purrr"
## [7] "readr" "tidyr" "tibble" "ggplot2" "tidyverse" "stats"
## [13] "graphics" "grDevices" "utils" "datasets" "methods" "base"
##
## [[3]]
## [1] "modelsummary" "fst" "lubridate" "forcats" "stringr"
## [6] "dplyr" "purrr" "readr" "tidyr" "tibble"
## [11] "ggplot2" "tidyverse" "stats" "graphics" "grDevices"
## [16] "utils" "datasets" "methods" "base"
##
## [[4]]
## [1] "viridis" "viridisLite" "modelsummary" "fst" "lubridate"
## [6] "forcats" "stringr" "dplyr" "purrr" "readr"
## [11] "tidyr" "tibble" "ggplot2" "tidyverse" "stats"
## [16] "graphics" "grDevices" "utils" "datasets" "methods"
## [21] "base"Task 1
Provide code and answer.
Prompt: in the tutorial, we calculated the average trust in others for France and visualized it. Using instead the variable ‘Trust in Parliament’ (trstplt) and the country of Spain (country file provided on course website), visualize the average trust by survey year. You can truncate the y-axis if you wish. Provide appropriate titles and labels given the changes. What are your main takeaways based on the visual (e.g., signs of increase, decrease, or stall)?
Loading Data into R
Let’s load our data. We will work with Spain.
spain_data <- read.fst("spain_data.fst")Reviewing & Adding
First, let’s calculate the average for a variable of interest and then visualize.
We will be working with a trust variable – trust in parliament, measured from 0-10, where 0 represents “No trust at all” and 10 is “Complete trust”.
Let’s check our work. One quick way to do so:
table(spain_data$trstplt) # if there are values that are not supposed to be there (e.g., 77, 88, 99 in this case), then we need to deal with it##
## 0 1 2 3 4 5 6 7 8 9 10 77 88 99
## 5165 1830 2329 2441 2085 2890 1154 639 355 80 71 46 336 31Before proceeding, we need to clean and transform our variable.
spain_data <- spain_data %>%
mutate(
trstplt = ifelse(trstplt %in% c(77, 88, 99), NA, trstplt), # set values 77, 88, and 99 to NA.
)table(spain_data$trstplt) # Now we deleted the unneccessary values##
## 0 1 2 3 4 5 6 7 8 9 10
## 5165 1830 2329 2441 2085 2890 1154 639 355 80 71Next, let’s create a ‘year’ variable from the essround variable. We will use it to visualize.
spain_data$year <- NA
replacements <- c(2002, 2004, 2006, 2008, 2010, 2012, 2014, 2016, 2018, 2020)
for(i in 1:10){
spain_data$year[spain_data$essround == i] <- replacements[i]
}Let’s check that it worked
table(spain_data$year)##
## 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020
## 1729 1663 1876 2576 1885 1889 1925 1958 1668 2283Now, we will calculate the average by year to then visualize:
trust_by_year <- spain_data %>%
group_by(year) %>%
summarize(mean_trust = mean(trstplt, na.rm = TRUE))
trust_by_year## # A tibble: 10 × 2
## year mean_trust
## <dbl> <dbl>
## 1 2002 3.41
## 2 2004 3.66
## 3 2006 3.49
## 4 2008 3.32
## 5 2010 2.72
## 6 2012 1.91
## 7 2014 2.23
## 8 2016 2.40
## 9 2018 2.55
## 10 2020 1.94We can see from this table that the average DOES shift much from year to year.
Now it’s time to:
Visualize
ggplot(trust_by_year, aes(x = year, y = mean_trust)) +
geom_line(color = "blue", size = 1) + # Line to show the trend
geom_point(color = "red", size = 3) + # Points to highlight each year's value
labs(title = "Trust in Others in Spain (2002-2020)",
x = "Survey Year",
y = "Average Trust (0-10 scale)") +
ylim(0, 10) + # Setting the y-axis limits from 0 to 10
theme_minimal()## Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
## ℹ Please use `linewidth` instead.
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.
Here we can see a general decrease in trust towards the Parliament year by year (only a slightly increase after 2015). After a period of relative stability until 2007, there’s a sharp decline until 2012. The trust levels reduced over the years.
Task 2
Provide answer only.
Prompt and question: Based on the figure we produced above called task2_plot, tell us: what are your main takeaways regarding France relative to Italy and Norway? Make sure to be concrete and highlight at least two important comparative trends visualized in the graph.
The proportion of “Saying Yes to Feeling Close to Political Party” of France stands relatively higher when we compared to the Italy. And lower than Norway.
Moreover, Norway has the highest proportion of saying yes to a party, whereas Italy has the least proportion among every cohorts each year and the proportion values for France is in the middle of amongst them in every cohort. Both of three countries’ proportion are decreasing year by year almost at the same rate of change.
Task 3
Provide code and answer.
Question: What is the marginal percentage of Italian men who feel close to a particular political party?
clsprty
Loading Data into R
Let’s load our data. We will work with Spain.
italy_data <- read.fst("italy_data.fst")italy_data <- italy_data %>%
mutate(
gndr = case_when(
gndr == 1 ~ "Male",
gndr == 2 ~ "Female",
TRUE ~ NA_character_
),
clsprty = case_when(
clsprty %in% 1 ~ "Yes",
clsprty %in% 2 ~ "No",
TRUE ~ NA_character_
)
) We wanted to showcase how to display the percentages of two categories (yes and no to a political party) by two other categories (male and female).
clsprty_percentages <- italy_data %>%
filter(!is.na(clsprty), !is.na(gndr)) %>%
group_by(gndr, clsprty) %>%
summarise(count = n(), .groups = 'drop') %>%
mutate(percentage = count / sum(count) * 100)
clsprty_percentages ## # A tibble: 4 × 4
## gndr clsprty count percentage
## <chr> <chr> <int> <dbl>
## 1 Female No 3228 34.2
## 2 Female Yes 1686 17.9
## 3 Male No 2593 27.5
## 4 Male Yes 1936 20.5Here we can see that the marginal percentage of Italian men (male) who feel close (says yes) to a particular political party is 20.502%.
Task 4
Provide code and output only.
Prompt: In the tutorial, we calculated then visualized the percentage distribution for left vs. right by gender for France. Your task is to replicate the second version of the visualization but for the country of Sweden instead.
Loading Data into R
Let’s load our data. We will work with Sweden now
sweden_data <- read.fst("sweden_data.fst")Before proceeding, we need to clean and transform our variable.
sweden_data <- sweden_data %>%
mutate(
lrscale = ifelse(lrscale %in% c(77, 88, 99), NA, lrscale), # set values 77, 88, and 99 to NA.
)table(sweden_data$lrscale) # Now we deleted the unneccessary values##
## 0 1 2 3 4 5 6 7 8 9 10
## 673 427 1193 2069 1826 3880 1765 2656 1905 493 587Percentages
sweden_data <- sweden_data %>%
mutate(
gndr = case_when(
gndr == 1 ~ "Male",
gndr == 2 ~ "Female",
TRUE ~ NA_character_ # Set anything that is not 1 or 2 to NA
),
lrscale = case_when(
lrscale %in% 0:3 ~ "Left", # Left-wing (0 to 3)
lrscale %in% 7:10 ~ "Right", # Right-wing (7 to 10)
TRUE ~ NA_character_ # Moderate (4, 5, 6) and special codes (77, 88, 99) set to NA
)
) Calculations
lrscale_percentages <- sweden_data %>% # Begin with the dataset 'sweden_data'
filter(!is.na(lrscale), !is.na(gndr)) %>% # Filter out rows where 'lrscale' or 'gender' is NA (missing data)
group_by(gndr, lrscale) %>% # Group the data by 'gender' and 'lrscale' categories
summarise(count = n(), .groups = 'drop') %>% # Summarise each group to get counts, and then drop groupings
mutate(percentage = count / sum(count) * 100) # Calculate percentage for each group by dividing count by total count and multiplying by 100
lrscale_percentages # The resulting dataframe## # A tibble: 4 × 4
## gndr lrscale count percentage
## <chr> <chr> <int> <dbl>
## 1 Female Left 2296 23.0
## 2 Female Right 2530 25.3
## 3 Male Left 2062 20.6
## 4 Male Right 3107 31.1Visualization
lrscale_plot <- ggplot(lrscale_percentages, aes(x = lrscale, y = percentage, fill = lrscale)) +
geom_bar(stat = "identity", position = position_dodge()) + # Dodged bar chart
facet_wrap(~ gndr, scales = "fixed") + # Fixed scales for y-axis across facets
scale_fill_brewer(palette = "Set1") + # Distinct colors for Left and Right
labs(
title = "Political Orientation (Left vs. Right) by Gender in Sweden",
x = "Political Orientation",
y = "Percentage of Respondents",
fill = "Orientation"
) +
theme_minimal() + # Minimal theme for clarity
theme(legend.position = "bottom") # Legend at the bottom
# Display the ggplot object
lrscale_plot
Task 5
Provide code and answer: In Hungary, what is the conditional probability of NOT feeling close to any particular party given that the person lives in a rural area?
Loading Data into R
Let’s load our data. We will work with Hungary.
hungary_data <- read.fst("hungary_data.fst")Reviewing & Adding
We will be working with the “probability of feeling close to any particular party” variable. From there, we’ll endevour to come to the “Feeling NOT close to a political party”
Let’s check our work. One quick way to do so:
table(hungary_data$clsprty) # if there are values that are not supposed to be there (e.g., 77, 88, 99 in this case), then we need to deal with it##
## 1 2 7 8 9
## 7342 8679 322 291 8So there is no need us to clean or transform our variable.
# Recode clsprty and geo variables, removing NAs
hungary_data <- hungary_data %>%
mutate(
geo = recode(as.character(domicil),
'1' = "Urban",
'2' = "Urban",
'3' = "Rural",
'4' = "Rural",
'5' = "Rural",
'7' = NA_character_,
'8' = NA_character_,
'9' = NA_character_)
) %>%
filter(!is.na(clsprty), !is.na(geo)) # Removing rows with NA in clsprty or geohungary_data <- hungary_data %>%
filter(!is.na(clsprty)) %>%
mutate(
clsprty = case_when(
clsprty == 1 ~ "Yes",
clsprty == 2 ~ "No"
)
) %>%
filter(!is.na(clsprty)) In the next step, we will calculate conditional probabilities where:
count(clsprty, geo) tallies the number of occurrences for each combination of clsprty and geo.
group_by(geo) groups the data by the geographical area.
mutate(prob = n / sum(n)) calculates the probability of each clsprty value given each geographical area.
# Calculate conditional probabilities, excluding NAs
cond <- hungary_data %>%
count(clsprty, geo) %>%
group_by(geo) %>%
mutate(prob = n / sum(n))
cond## # A tibble: 4 × 4
## # Groups: geo [2]
## clsprty geo n prob
## <chr> <chr> <int> <dbl>
## 1 No Rural 6275 0.554
## 2 No Urban 2395 0.512
## 3 Yes Rural 5055 0.446
## 4 Yes Urban 2283 0.488So the conditional probability of NOT feeling close to any particular party given that the person lives in a rural area is 0.554