PA 8: United Nations Voting Records

Author

Instructions

This task is complex. It requires many different types of abilities. Everyone will be good at some of these abilities but nobody will be good at all of them. In order to solve this puzzle, you will need to use the skills of each member of your group.

Groupwork Protocols

During the Practice Activity, you and your partner will alternate between two roles—Computer and Coder.

When you are the Computer, you will type into the Quarto document in RStudio. However, you do not type your own ideas. Instead, you type what the Coder tells you to type. You are permitted to ask the Coder clarifying questions, and, if both of you have a question, you are permitted to ask the professor. You are expected to run the code provided by the Coder and, if necessary, to work with the Coder to debug the code. Once the code runs, you are expected to collaborate with the Coder to write code comments that describe the actions taken by your code.

When you are the Coder, you are responsible for reading the instructions / prompts and directing the Computer what to type in the Quarto document. You are responsible for managing the resources your group has available to you (e.g., cheatsheet, textbook). If necessary, you should work with the Computer to debug the code you specified. Once the code runs, you are expected to collaborate with the Computer to write code comments that describe the actions taken by your code.

Here are more details of the Pair Programming Protocols

Note

The partner was born the furthest away from CSUMB will start as the Computer (typing and listening to instructions from the Coder).

Group Norms

Remember, your group is expected to adhere to the following norms:

Be curious. Don’t correct.
Be open minded.
Ask questions rather than contribute.
Respect each other.
Allow each teammate to contribute to the activity through their role.
Do not divide the work.
No cross talk with other groups.
Communicate with each other!

Goals for the Activity

Join multiple data tables together by a common variable(s)
Create new data sets through the joining of data from various sources
Combine join functions with other tidyverse functions

THROUGHOUT THE Activity be sure to follow the Style Guide by doing the following:

load the appropriate packages at the beginning of the Quarto document
use proper spacing
add labels to all code chunks
comment at least once in each code chunk to describe why you made your coding decisions
add appropriate labels to all graphic axes

Setting up your Project

Your project should have the following components:

completed pa-8-united-nations-voting-activity.qmd
rendered file as .html

Important

The original Computer should submit the zip file for the canvas quiz. The original Coder can just submit the rendered html file.

Computer - Be sure to share the final .qmd and .html file with the original Coder.

Data Description

The data this week comes from Harvard’s Dataverse by way of Mine Çetinkaya-Rundel, David Robinson, and Nicholas Goguen-Compagnoni.

Original Data citation: Erik Voeten “Data and Analyses of Voting in the UN General Assembly” Routledge Handbook of International Organization, edited by Bob Reinalda (published May 27, 2013). Available at SSRN: http://ssrn.com/abstract=2111149

It was featured on TidyTuesday

Here is each data set and its description (you might want to look at the Tidy Tuesday link for the tables already rendered)

unvotes.csv

variable	class	description
rcid	double	The roll call id
country	character	Country name, by official English short name
country_code	character	2-character ISO country code
vote	integer	Vote result as a factor of yes/abstain/no

rcid	country	country_code	vote
unvotes
4501	Turkmenistan	TM	yes
2342	Saudi Arabia	SA	yes
3952	Moldova	MD	no
3612	Romania	RO	yes
2481	German Democratic Republic	DD	no
4543	Greece	GR	no
1351	United States	US	no
5073	Indonesia	ID	yes
4397	Kyrgyzstan	KG	yes
3194	Samoa	WS	yes

roll_calls.csv

variable	class	description
rcid	integer	The roll call id
session	double	Session number. The UN holds one session per year; these started in 1946
importantvote	integer	Whether the vote was classified as important by the U.S. State Department report “Voting Practices in the United Nations”. These classifications began with session 39
date	double	Date of the vote, as a Date vector
unres	character	Resolution code
amend	integer	Whether the vote was on an amendment; coded only until 1985
para	integer	Whether the vote was only on a paragraph and not a resolution; coded only until 1985
short	character	Short description
descr	character	Longer description

rcid	session	importantvote	date	unres	amend	para	short	descr
roll_calls
1467	28	0	1973-12-03	R/28/3117	0	0	INDEPENDENCE, COLONIAL COUNTRIES	RES, REITERATING REQ SA, ESP UNDP & IBRD, INCREASE ASSISTANCE TO REFUGEES FROM COLONIAL TERRS; WITHHOLD ASSIST FROM GOVERNMENTS OF PORT, SO. AFR & SO. RHODESIA UNTIL THEY RENOUNCE DISCRIM POLICIES; REQ COOPERATION WITH OAU; REQ ECOSOC CONSIDER
1302	26	0	1971-12-07	R/26/2901	0	0	BUDGET ESTIMATES, 1972	TO ADOPT THE RESOL., I.A., ESTAB WORKING CAPITAL FUND FOR FINANCIAL YEAR 1972 IN AMOUNT OF $40 MILLION, AND DESCRIBING MANNER MEMBER STATES SHALL MAKE ADVANCES TO FUND AS WELL AS AUTHORIZING SYG TO MAKE ADVANCES FROM THE FUND
2374	37	0	1982-12-01	R/37/74A	0	0	DENUCLEARIZATION, AFRICA	TO ONCE AGAIN REITERATE ITS CALL UPON ALL STATES TO CONSIDER AND RESPECT THE CONTINENT OF AFRICA AND ITS SURROUNDING AREAS AS A NUCLEAR-WEAPON-FREE, AND TO CONDEMN ALL FORMS OF NUCLEAR COLLABORATION BY ANY STATE, CORPORATION, INSTITUTION OR IND
4213	54	0	1999-12-02	R/54/84	NA	NA	SELF-DETERMINATION, ECONOMIC IMPEDIMENTS	Economic and other activities which affect the interests of the peoples of the Non-Self-Governing Territories
3206	42	0	1987-12-02	R/42/149	NA	NA	INTERNATIONAL ECONOMIC ORDER	Progressive development of the principles and norms of international law relating to the new international economic order
1289	26	0	1971-12-07	R/26/2889	0	0	BUDGET ESTIMATES, 1972	TO ADOPT THE RESOL., I.A., DECIDING THAT EFFECTIVE 1 JANUARY 1972, HONORARIUM OF CHAIRMAN OF ACABQ SHALL BE $25,000 PER ANNUM, PROVIDED HE IS NOT ACTIVELY ENGAGED ON BEHALF OF HIS GOVT OR ANOTHER BODY.
5618	72	0	2017-12-04	A/RES/72/25	NA	NA	Conclusion of effective international arrangements to assure non-nuclear-weapon States against the use or threat of use of nuclear weapons : resolution / adopted by the General Assembly	A/72/251 96 - Conclusion of effective international arrangements to assure non-nuclear-weapon States against the use or threat of use of nuclear weapons. - NON-NUCLEAR-WEAPON STATES--SECURITY
5631	72	0	2017-12-07	A/RES/72/87	NA	NA	Israeli practices affecting the human rights of the Palestinian people in the Occupied Palestinian Territory, including East Jerusalem : resolution / adopted by the General Assembly	A/72/251 54 - Report of the Special Committee to Investigate Israeli Practices Affecting the Human Rights of the Palestinian People and Other Arabs of the Occupied Territories. - TERRITORIES OCCUPIED BY ISRAEL--HUMAN RIGHTS
5039	64	0	2009-12-21	R/64/188	NA	NA	INTERNATIONAL TRADE	International trade and development : resolution / adopted by the General Assembly
4424	57	1	2002-12-05	R/57/190	NA	NA	RIGHTS OF THE CHILD	Rights of the child

issues.csv

variable	class	description
rcid	integer	The roll call id
short_name	character	Two-letter issue codes
issue	integer	Descriptive issue name

rcid	short_name	issue
issues
2065	me	Palestinian conflict
2527	nu	Nuclear weapons and nuclear material
4795	co	Colonialism
2615	hr	Human rights
5849	di	Arms control and disarmament
4804	hr	Human rights
979	co	Colonialism
2357	me	Palestinian conflict
1684	nu	Nuclear weapons and nuclear material
4722	co	Colonialism

What variable(s) do each of these data frames have in common?

Data Exploration

Our goal today is to explore how various members (countries) in the United Nations vote. We have three data sets, what can we determine from each data set separately?

UN Votes

The first data set, unvotes contains data on the rcid which is the roll call id for the vote, the country/country code, and how the country voted. What can we learn from the data?

Comment on the following code, what is happening in each line? One way to approach seeing what each line does is to highlight the code from before the pipe of that line up to the data unvotes and use CTRL + ENTER to run just the highlighted lines.

unvotes |> 
  count(country, vote) |> #comment
  group_by(country) |> #comment
  mutate(total = sum(n)) |> #comment 
  mutate(prop_vote = n/total) |> #comment
  filter(country %in% c("United States", "Canada", 
                        "Germany", "France",
                        "Italy", "Japan",
                        "United Kingdom")) |> #comment
  ggplot(aes(y = country, x = prop_vote, 
             fill = vote)) + #comment
  geom_col(position = position_stack()) + #comment
  labs(y = "Group of Seven Countries",
       x = "Proportion of Votes",
       title = "Voting Record of the G7",
       fill = "Vote") + #comment
  theme_minimal() +  #comment
  scale_fill_viridis_d(end = 0.8)

Describe what the graph above demonstrates above UN voting records for the G7.

Roll Calls

The second data set, roll_calls has more information on the type of vote, the importance, whether it was a resolution, and date of the vote.

roll_calls |> 
  distinct(short)

# A tibble: 2,019 × 1
   short                             
   <chr>                             
 1 AMENDMENTS, RULES OF PROCEDURE    
 2 SECURITY COUNCIL ELECTIONS        
 3 VOTING PROCEDURE                  
 4 DECLARATION OF HUMAN RIGHTS       
 5 GENERAL ASSEMBLY ELECTIONS        
 6 ECOSOC POWERS                     
 7 POST-WAR RECONSTRUCTION           
 8 U.N. MEMBERS, RELATIONS WITH SPAIN
 9 TRUSTEESHIP AMENDMENTS            
10 COUNCIL MEMBER TERM LENGTH        
# ℹ 2,009 more rows

What does the code above do? What information does it provide? Is it useful? > Description of code results

We can use the individual data for roll_calls to look at the number of votes per year over time.

roll_calls |> 
  mutate(year = lubridate::year(date)) |> #extracts the year from the date value and creates a new `year` column
  count(year) |> #counts how many votes there were per year assuming each line is an single voting instance
  ggplot(aes(x = year, y = n)) +
  geom_line() +
  labs(x = "Year", y = "Number of Votes",
       title = "UN Votes per Year") +
  theme_minimal()

What information is missing from the above graphic that might be useful in understanding the issues the UN commonly votes on?

Issues

Finally we have the issues data which provides a more general description for each vote on specific issues. Note that not all issues are included in the data set, just the ones related to the 6 issues below:

issues |> 
  count(issue) |> 
  adorn_totals("row") #from janitor package

                                issue    n
         Arms control and disarmament 1092
                          Colonialism  957
                 Economic development  765
                         Human rights 1015
 Nuclear weapons and nuclear material  855
                 Palestinian conflict 1061
                                Total 5745

Notice the size of the issues data - it has 5745 rows, but if look at the distinct number of roll call identification numbers, you will see that there are 4099, meaning that more than one issue can be associated with the same roll call id/vote.

issues |> 
  distinct(rcid) |> 
  n_distinct()

[1] 4099

Recall that in roll_call there are 6202 distinct roll call ids/votes, so the issues associated with issues do not represent all votes (i.e., there were other U.N. votes on other issues than our 6 chosen issues).

roll_calls |> 
  distinct(rcid) |> 
  n_distinct()

[1] 6202

It would be helpful to use the issues data with the roll_calls data to be able to better understand the voting trends within the UN on these 6 issues. To do this, we need to join the data.

Votes Over Time

Now let’s join our data together to get a better idea of how the UN has voted over time.
First, look at the number of rows in issues and roll_calls - do they match? What does this indicate?

dim(issues)

[1] 5745    3

dim(roll_calls)

[1] 6202    9

Now let’s try joining the roll_calls with the issues data. Compare the following codes to join the data. Describe what each one does and how it differs from the others as a comment in the code chunk. You might need to reference the slides or reading for this week.

roll_calls |> 
  left_join(issues, by = "rcid") #description of join

roll_calls |> 
  right_join(issues, by = "rcid") #description of join

roll_calls |> 
  full_join(issues, by = "rcid") #description of join

roll_calls |> 
  inner_join(issues, by = "rcid") #description of join

If we are only interested in retaining the records associated with the issues labeled in that data frame, ignoring the other votes, which join should we use?

Now that we know how to join the data, we will use the following code to examine the the voting trends for three of the issues related to conflict/weapons.

Be sure to run the code via the green arrow on the code chunk, as the case_when() code can get finicky sometimes and claim an error about a comma in the code when it doesn’t exist. Comment on the code where indicated and add your chosen join function

roll_calls |> 
 __________join(issues, by = "rcid") |>  #join roll_calls and issues so that just the votes related to the issues data are retained.
  mutate(issue_short = case_when(
    issue == "Arms control and disarmament" ~ "Arms Control",
    issue == "Nuclear weapons and nuclear material" ~ "Nuclear Weapons",
    issue == "Palestinian conflict" ~ "Palestinian Conflict",
    TRUE ~ issue)) |>   #what is happening in this mutate function?
  filter(issue_short %in% c("Arms Control",
                            "Nuclear Weapons",
                            "Palestinian Conflict")) |> #what are we doing here?
  mutate(year = lubridate::year(date)) |> #create a column `year` that contains the year value
  count(year, 
        issue_short) |> #what does this line do?
  ggplot(aes(x = year, 
             y = n, 
             group = issue_short)) + #what does this line do?
  geom_line(aes(color = issue_short)) + #what does this line do?
  labs(x = "Year", 
       y = "Number of Votes",
       title = "United Nations Votes per Year",
       subtitle = "Conflict and Arms Related Votes",
       color = "Voting Issue") +
  theme_minimal() +
  scale_color_viridis_d(end = 0.9)

What do you notice? What do you wonder based on the graph created?

Joining all Data

We want to try create a visualization that compares the voting records of the US and Canada on the 6 major issues of interest. To do this, though, we need information from all three data set, unvotes, issues, and roll_calls

We want to join all three data sets together, maintaining only the votes for which we have identified the general issue (e.g., Nuclear War, Arms, Economics, etc.), but recognizing that each rcid will match MULTIPLE rows in the unvotes because we have each individual country’s vote. We will save (assign) the data as un_full.

un_full <- roll_calls |> 
  right_join(issues, by = "rcid") |> 
  left_join(unvotes, by = "rcid", 
            multiple = "all", 
            relationship = "many-to-many")

Describe what each join is doing and why each join has specific arguments.

Note

It will be helpful to look up the arguments in the left_join() function on the dplyr webpage.

Now, we are going to do some data cleaning. Our goal is create a data set that includes the percentage of “yes” votes per country each year. We will call the data table yes_votes. Provide a comment to describe what each line of code is doing in the process.

yes_votes <- un_full |> 
  select(country, 
         issue, 
         date, 
         vote) |>  # your comment here
  mutate(year = lubridate::year(date)) |> #create a new variable called year
  group_by(country, 
           year, 
           issue) |> # your comment here
  summarize(prop_yes = mean(vote == "yes"), 
            .groups = "drop_last") |> #calculate the proportion of yes votes
  mutate(issue = case_when(
    issue == "Arms control and disarmament" ~ "Arms Control",
    issue == "Nuclear weapons and nuclear material" ~ "Nuclear Weapons",
    issue == "Palestinian conflict" ~ "Palestinian Conflict",
    TRUE ~ issue)) # your comment here

Now we can feed the yes_votes transformed data table into your graphing code, but first we will want to focus on the United States and Canada. Provide a comment to describe what each line of code is doing in the process.

yes_votes |> 
  filter(country %in% c("United States","Canada")) |> # your comment here
  ggplot(mapping = aes(x = year, y = prop_yes, color = country)) + # your comment here
  geom_point(alpha = 0.4) + # your comment here
  #geom_line(aes(group = country)) +
  geom_smooth(method = "loess", se = FALSE) + #this fits a special model called a loess regression, a smooth line that fits the data
  facet_wrap(~issue) + # your comment here
  scale_y_continuous(labels = scales::percent) + #your comment here
  labs(
    title = "Percentage of 'Yes' votes in the UN General Assembly",
    subtitle = "1946 to 2019",
    y = "% Yes",
    x = "Year",
    color = "Country"
  ) + #your comment here
  theme_bw() +
  theme(legend.position = "bottom") + #your comment here 
  scale_color_viridis_d(option = "turbo") #your comment here

What do you notice about the voting records over time?

Demonstration: Adding more data

After your instructor created the above plot, she became curious about how politics might impact the UN Voting record for the United States since UN Ambassador is a presidential appointment. So your instructor started searching for a data set of US presidents, their years in office, and their political affiliation. She found a data set on Kaggle.com and removed the information prior to 1940 (because the dates were coded funny and it was causing problems). She saved the data as us_presidents.csv and imported it into the project.

president <- read_csv("data/us_presidents.csv", show_col_types = FALSE)
slice_sample(president, n = 10) |>   
  gt() |> 
  tab_header(title = "president")

id	S.No.	start	end	president	prior	party	vice
president
40	41	1/20/1989	20-Jan-93	George H. W. Bush	43rd Vice President of the United States	Republican	Dan Quayle
32	33	4/12/1945	20-Jan-53	Harry S. Truman	34th Vice President of the United States	Democratic	Office vacant
35	36	11/22/1963	20-Jan-69	Lyndon B. Johnson	37th Vice President of the United States	Democratic	Office vacant
29	30	8/2/1923	4-Mar-29	Calvin Coolidge	29th Vice President of the United States	Republican	Office vacant
36	37	1/20/1969	9-Aug-74	Richard Nixon	36th Vice President of the United States (1953–1961)	Republican	Spiro Agnew
38	39	1/20/1977	20-Jan-81	Jimmy Carter	76th Governor of Georgia (1971–1975)	Democratic	Walter Mondale
34	35	1/20/1961	22-Nov-63	John F. Kennedy	U.S. Senator ( Class 1 ) from Massachusetts (1953–1960)	Democratic	Lyndon B. Johnson
26	27	3/4/1909	4-Mar-13	William Howard Taft	42nd United States Secretary of War (1904–1908)	Republican	James S. Sherman
39	40	1/20/1981	20-Jan-89	Ronald Reagan	33rd Governor of California ( 1967–1975 )	Republican	George H. W. Bush
42	43	1/20/2001	20-Jan-09	George W. Bush	46th Governor of Texas ( 1995–2000 )	Republican	Dick Cheney

She realized that her data only had the start/end dates for each president and she wanted a data set that filled in the missing years and political party for the president in that time period. After much googling and reading Stack Overflow, found two functions she did not know about called complete() and fill() to fill in the missing years and party affiliations

politics_year <- president |> 
    mutate(start = lubridate::mdy(start),  #formats date correctly
           start_year = lubridate::year(start)) |> #pulls out year
  filter(start_year > 1940) |>  #removes data before 1940 since there was no UN
  select(start_year, party) |>   #pulls out just the variables of interest
  complete(start_year = seq(min(start_year), 2020, by = 1)) |> #fill in missing years
  fill(party) #fill in missing party affiliations for years

slice_tail(politics_year, n = 10) |>   
  gt() |> 
  tab_header(title = "Year by Presidential Party")

start_year	party
Year by Presidential Party
2011	Democratic
2012	Democratic
2013	Democratic
2014	Democratic
2015	Democratic
2016	Democratic
2017	Republican
2018	Republican
2019	Republican
2020	Republican

Next, your instructor took the yes_votes data and filtered out just the United States data and then joined by year to add the party affiliation of the president for each year of UN votes. To create the visualization with the smoothed model, but color by party affiliation, she had to add a new column called predict that fit the model first instead of using geom_smooth().

yes_votes |> 
  filter(country == "United States") |>  #we want to focus on the US
  left_join(politics_year, 
            by = c("year" = "start_year")) |> #adding in president political affiliation data
  group_by(issue) |>  #we want to calculate predicted yes by issue type 
  mutate(predict = predict(loess(prop_yes ~ year))) |>  #creates predictions using a loess model for yes vs year
  ungroup() |> #ungroups the group_by line so it doesn't mess with other calculations
  ggplot(mapping = aes(x = year, 
                       color = party, 
                       group = 1)) + #sets universal aes to observe changes over time
  geom_point(aes(y = prop_yes),
             alpha = 0.4) + #plots points for proportion yes but transparent
  geom_line(aes(y = predict)) + #plots the predicted yeses for votes
  facet_wrap(~issue) + #creates a plot for each issue
  scale_y_continuous(labels = scales::percent) + #scales y-axis as percent values
  labs(
    title = "Percentage of United States 'Yes' votes in the UN General Assembly",
    subtitle = "1946 to 2019",
    y = "% Yes",
    x = "Year",
    color = "Presidential Political Party"
  ) + #adds labels
  theme_bw() + #changes appearance
  scale_color_manual(values = c("blue", "red"),
                     breaks = c("Democratic", "Republican")) + #sets specific colors for each party
  theme(legend.position = "bottom")  #moves legend to bottom to create space

--- title: "PA 8: United Nations Voting Records" author: "Instructions" format: html code-tools: true toc: true editor: source execute: echo: true --- ```{r} #| label: setup #| include: false library(tidyverse) library(lubridate) #should already be installed as a part of the tidyverse, we will talk more about this in Week 10 library(scales) #you may need to install package `scales` library(gt) #for tables library(janitor) ``` ***This task is complex. It requires many different types of abilities. Everyone will be good at some of these abilities but nobody will be good at all of them. In order to solve this puzzle, you will need to use the skills of each member of your group.*** # Groupwork Protocols During the Practice Activity, you and your partner will alternate between two roles---Computer and Coder. When you are the **Computer**, you will type into the Quarto document in RStudio. However, you **do not** type your own ideas. Instead, you type what the Coder tells you to type. You are permitted to ask the Coder clarifying questions, and, if both of you have a question, you are permitted to ask the professor. You are expected to run the code provided by the Coder and, if necessary, to work with the Coder to debug the code. Once the code runs, you are expected to collaborate with the Coder to write code comments that describe the actions taken by your code. When you are the **Coder**, you are responsible for reading the instructions / prompts and directing the Computer what to type in the Quarto document. You are responsible for managing the resources your group has available to you (e.g., cheatsheet, textbook). If necessary, you should work with the Computer to debug the code you specified. Once the code runs, you are expected to collaborate with the Computer to write code comments that describe the actions taken by your code. Here are more details of the [Pair Programming Protocols](../pair-programming-norms.qmd) ::: {.callout-note} The partner was born the furthest away from CSUMB will start as the Computer (typing and listening to instructions from the Coder). ::: ### Group Norms Remember, your group is expected to adhere to the following norms: 1. Be curious. Don't correct. 2. Be open minded. 3. Ask questions rather than contribute. 4. Respect each other. 5. Allow each teammate to contribute to the activity through their role. 6. Do not divide the work. 7. No cross talk with other groups. 8. Communicate with each other! ## Goals for the Activity - Join multiple data tables together by a common variable(s) - Create new data sets through the joining of data from various sources - Combine `join` functions with other `tidyverse` functions **THROUGHOUT THE Activity** be sure to follow the Style Guide by doing the following: - load the appropriate packages at the beginning of the Quarto document - use proper spacing - *add labels* to all code chunks - comment at least once in each code chunk to describe why you made your coding decisions - add appropriate labels to all graphic axes ## Setting up your Project Your project should have the following components: 1. completed [pa-8-united-nations-voting-activity.qmd](../week-9/pa-8-united-nations-voting-activity.qmd) 2. rendered file as `.html` ::: {.callout-important} The original `Computer` should submit the zip file for the canvas quiz. The original `Coder` can just submit the rendered html file. `Computer` - Be sure to share the final .qmd and .html file with the original `Coder`. ::: # Data Description The data this week comes from Harvard's Dataverse by way of Mine Çetinkaya-Rundel, David Robinson, and Nicholas Goguen-Compagnoni. > Original Data citation: Erik Voeten "Data and Analyses of Voting in the UN General Assembly" Routledge Handbook of International Organization, edited by Bob Reinalda (published May 27, 2013). Available at SSRN: http://ssrn.com/abstract=2111149 It was featured on [TidyTuesday](https://github.com/rfordatascience/tidytuesday/blob/master/data/2021/2021-03-23/readme.md) Here is each data set and its description (you might want to look at the Tidy Tuesday link for the tables already rendered) `unvotes.csv` |variable |class |description | |:------------|:---------|:-----------| |rcid |double | The roll call id | |country |character | Country name, by official English short name | |country_code |character | 2-character ISO country code | |vote |integer | Vote result as a factor of yes/abstain/no | ```{r} #| label: read-unvotes #| echo: false unvotes <- read_csv("data/unvotes.csv", show_col_types = FALSE) slice_sample(unvotes, n = 10) |> gt() |> tab_header(title = "unvotes") ``` `roll_calls.csv` |variable |class |description | |:-------------|:---------|:-----------| |rcid |integer |The roll call id | |session |double | Session number. The UN holds one session per year; these started in 1946| |importantvote |integer | Whether the vote was classified as important by the U.S. State Department report "Voting Practices in the United Nations". These classifications began with session 39| |date |double | Date of the vote, as a Date vector| |unres |character | Resolution code | |amend |integer | Whether the vote was on an amendment; coded only until 1985 | |para |integer | Whether the vote was only on a paragraph and not a resolution; coded only until 1985| |short |character | Short description | |descr |character | Longer description| ```{r} #| label: read-roll_calls #| echo: false #edit to read in roll_calls roll_calls <- read_csv("data/roll_calls.csv", show_col_types = FALSE) slice_sample(roll_calls, n = 10) |> gt() |> tab_header(title = "roll_calls") ``` `issues.csv` |variable |class |description | |:----------|:---------|:-----------| |rcid |integer | The roll call id| |short_name |character | Two-letter issue codes | |issue |integer | Descriptive issue name | ```{r} #| label: read-issues #| echo: false #edit to read in issues issues <- read_csv("data/issues.csv", show_col_types = FALSE) slice_sample(issues, n = 10) |> gt() |> tab_header(title = "issues") ``` **What variable(s) do each of these data frames have in common?**  ## Data Exploration Our goal today is to explore how various members (countries) in the United Nations vote. We have three data sets, what can we determine from each data set separately? ### UN Votes The first data set, `unvotes` contains data on the `rcid` which is the roll call id for the vote, the country/country code, and how the country voted. What can we learn from the data? Comment on the following code, what is happening in each line? One way to approach seeing what each line does is to highlight the code from before the `pipe` of that line up to the data `unvotes` and use `CTRL + ENTER` to run just the highlighted lines. ```{r} #| label: unvotes-graph unvotes |> count(country, vote) |> #comment group_by(country) |> #comment mutate(total = sum(n)) |> #comment mutate(prop_vote = n/total) |> #comment filter(country %in% c("United States", "Canada", "Germany", "France", "Italy", "Japan", "United Kingdom")) |> #comment ggplot(aes(y = country, x = prop_vote, fill = vote)) + #comment geom_col(position = position_stack()) + #comment labs(y = "Group of Seven Countries", x = "Proportion of Votes", title = "Voting Record of the G7", fill = "Vote") + #comment theme_minimal() + #comment scale_fill_viridis_d(end = 0.8) ``` **Describe what the graph above demonstrates above UN voting records for the G7.**  ### Roll Calls The second data set, `roll_calls` has more information on the type of vote, the importance, whether it was a resolution, and date of the vote. ```{r} #| label: roll-call-list roll_calls |> distinct(short) ``` **What does the code above do? What information does it provide? Is it useful?** > Description of code results We can use the individual data for `roll_calls` to look at the number of votes per year over time. ```{r} #| label: roll-call-graph roll_calls |> mutate(year = lubridate::year(date)) |> #extracts the year from the date value and creates a new `year` column count(year) |> #counts how many votes there were per year assuming each line is an single voting instance ggplot(aes(x = year, y = n)) + geom_line() + labs(x = "Year", y = "Number of Votes", title = "UN Votes per Year") + theme_minimal() ``` **What information is missing from the above graphic that might be useful in understanding the issues the UN commonly votes on?**  ### Issues Finally we have the `issues` data which provides a more general description for each vote on specific issues. Note that not all issues are included in the data set, just the ones related to the 6 issues below: ```{r} #| label: count-issues issues |> count(issue) |> adorn_totals("row") #from janitor package ``` Notice the size of the issues data - it has 5745 rows, but if look at the distinct number of roll call identification numbers, you will see that there are 4099, meaning that more than one issue can be associated with the same roll call id/vote. ```{r} #| label: distinct-issues-rcid issues |> distinct(rcid) |> n_distinct() ``` Recall that in `roll_call` there are 6202 distinct roll call ids/votes, so the issues associated with `issues` do not represent all votes (i.e., there were other U.N. votes on other issues than our 6 chosen issues). ```{r} #| label: distinct-issues-rcid-n roll_calls |> distinct(rcid) |> n_distinct() ``` It would be helpful to use the `issues` data with the `roll_calls` data to be able to better understand the voting trends within the UN on these 6 issues. To do this, we need to join the data. # Votes Over Time Now let's join our data together to get a better idea of how the UN has voted over time. First, look at the number of rows in `issues` and `roll_calls` - do they match? What does this indicate? ```{r} #| label: check-dimensions dim(issues) dim(roll_calls) ``` Now let's try joining the `roll_calls` with the `issues` data. Compare the following codes to join the data. Describe what each one does and how it differs from the others as a comment in the code chunk. You might need to reference the slides or reading for this week. ```{r} #| label: left-join #| message: true #| warning: true #| eval: false roll_calls |> left_join(issues, by = "rcid") #description of join ``` ```{r} #| label: right-join #| message: true #| warning: true #| eval: false roll_calls |> right_join(issues, by = "rcid") #description of join ``` ```{r} #| label: full-join #| message: true #| warning: true #| eval: false roll_calls |> full_join(issues, by = "rcid") #description of join ``` ```{r} #| label: inner-join #| message: true #| warning: true #| eval: false roll_calls |> inner_join(issues, by = "rcid") #description of join ``` **If we are only interested in retaining the records associated with the `issues` labeled in that data frame, ignoring the other votes, which join should we use?**  Now that we know how to join the data, we will use the following code to examine the the voting trends for three of the issues related to conflict/weapons. Be sure to run the code via the green arrow on the code chunk, as the `case_when()` code can get finicky sometimes and claim an error about a comma in the code when it doesn't exist. *Comment on the code where indicated* and add your chosen join function ```{r} #| label: votes-on-conflicts-graph #| eval: false roll_calls |> __________join(issues, by = "rcid") |> #join roll_calls and issues so that just the votes related to the issues data are retained. mutate(issue_short = case_when( issue == "Arms control and disarmament" ~ "Arms Control", issue == "Nuclear weapons and nuclear material" ~ "Nuclear Weapons", issue == "Palestinian conflict" ~ "Palestinian Conflict", TRUE ~ issue)) |> #what is happening in this mutate function? filter(issue_short %in% c("Arms Control", "Nuclear Weapons", "Palestinian Conflict")) |> #what are we doing here? mutate(year = lubridate::year(date)) |> #create a column `year` that contains the year value count(year, issue_short) |> #what does this line do? ggplot(aes(x = year, y = n, group = issue_short)) + #what does this line do? geom_line(aes(color = issue_short)) + #what does this line do? labs(x = "Year", y = "Number of Votes", title = "United Nations Votes per Year", subtitle = "Conflict and Arms Related Votes", color = "Voting Issue") + theme_minimal() + scale_color_viridis_d(end = 0.9) ``` ```{r} #| label: votes-on-conflicts-graph-correct #| echo: false roll_calls |> right_join(issues, by = "rcid") |> #join roll_calls and issues so that just the votes related to the issues data are retained. mutate(issue_short = case_when( issue == "Arms control and disarmament" ~ "Arms Control", issue == "Nuclear weapons and nuclear material" ~ "Nuclear Weapons", issue == "Palestinian conflict" ~ "Palestinian Conflict", TRUE ~ issue)) |> #simplifies the issue code for three issues filter(issue_short %in% c("Arms Control", "Nuclear Weapons", "Palestinian Conflict")) |> #filters the 3 codes of interest mutate(year = lubridate::year(date)) |> #create a column `year` that contains the year value count(year, issue_short) |> #counts up the issue votes by year? ggplot(aes(x = year, y = n, group = issue_short)) + #connects the lines by issue code geom_line(aes(color = issue_short)) + #colors lines by issue code labs(x = "Year", y = "Number of Votes", title = "United Nations Votes per Year", subtitle = "Conflict and Arms Related Votes", color = "Voting Issue") + theme_minimal() + scale_color_viridis_d(end = 0.9) ``` **What do you notice? What do you wonder based on the graph created?**  # Joining all Data We want to try create a visualization that compares the voting records of the US and Canada on the 6 major issues of interest. To do this, though, we need information from all three data set, `unvotes`, `issues`, and `roll_calls` We want to join all three data sets together, *maintaining only the votes for which we have identified the general issue* (e.g., Nuclear War, Arms, Economics, etc.), but recognizing that each `rcid` will match MULTIPLE rows in the `unvotes` because we have each individual country's vote. We will save (assign) the data as `un_full`. ```{r} #| label: join-all un_full <- roll_calls |> right_join(issues, by = "rcid") |> left_join(unvotes, by = "rcid", multiple = "all", relationship = "many-to-many") ``` **Describe what each join is doing and why each join has specific arguments.** ::: {.callout-note} It will be helpful to look up the arguments in the `left_join()` function on the [`dplyr` webpage](https://dplyr.tidyverse.org/reference/mutate-joins.html). ::: Now, we are going to do some data cleaning. Our goal is create a data set that includes the percentage of "yes" votes per country each year. We will call the data table `yes_votes`. *Provide a comment to describe what each line of code is doing in the process*. ```{r} #| label: yes-votes-cleaning yes_votes <- un_full |> select(country, issue, date, vote) |> # your comment here mutate(year = lubridate::year(date)) |> #create a new variable called year group_by(country, year, issue) |> # your comment here summarize(prop_yes = mean(vote == "yes"), .groups = "drop_last") |> #calculate the proportion of yes votes mutate(issue = case_when( issue == "Arms control and disarmament" ~ "Arms Control", issue == "Nuclear weapons and nuclear material" ~ "Nuclear Weapons", issue == "Palestinian conflict" ~ "Palestinian Conflict", TRUE ~ issue)) # your comment here ``` ```{r} #| label: yes-votes-cleaning-correct #| echo: false yes_votes <- un_full |> select(country, issue, date, vote) |> #keep just the columns for country, issue, date, vote mutate(year = lubridate::year(date)) |> #create a new variable called year group_by(country, year, issue) |> #group country, year, issue together to prepare for subsequent analysis summarize(prop_yes = mean(vote == "yes"), .groups = "drop_last") |> #calculate the proportion of yes votes mutate(issue = case_when( issue == "Arms control and disarmament" ~ "Arms Control", issue == "Nuclear weapons and nuclear material" ~ "Nuclear Weapons", issue == "Palestinian conflict" ~ "Palestinian Conflict", TRUE ~ issue)) #create/overwrite issue column with simpler values #View(yes_votes) #you can use this code to look at the table created ``` Now we can feed the `yes_votes` transformed data table into your graphing code, but first we will want to focus on the United States and Canada. *Provide a comment to describe what each line of code is doing in the process*. ```{r} #| label: yes-votes-canada-us #| message: false yes_votes |> filter(country %in% c("United States","Canada")) |> # your comment here ggplot(mapping = aes(x = year, y = prop_yes, color = country)) + # your comment here geom_point(alpha = 0.4) + # your comment here #geom_line(aes(group = country)) + geom_smooth(method = "loess", se = FALSE) + #this fits a special model called a loess regression, a smooth line that fits the data facet_wrap(~issue) + # your comment here scale_y_continuous(labels = scales::percent) + #your comment here labs( title = "Percentage of 'Yes' votes in the UN General Assembly", subtitle = "1946 to 2019", y = "% Yes", x = "Year", color = "Country" ) + #your comment here theme_bw() + theme(legend.position = "bottom") + #your comment here scale_color_viridis_d(option = "turbo") #your comment here ``` **What do you notice about the voting records over time?** # Demonstration: Adding more data After your instructor created the above plot, she became curious about how politics might impact the UN Voting record for the United States since UN Ambassador is a presidential appointment. So your instructor started searching for a data set of US presidents, their years in office, and their political affiliation. She found a data set on Kaggle.com and removed the information prior to 1940 (because the dates were coded funny and it was causing problems). She saved the data as `us_presidents.csv` and imported it into the project. ```{r} #| label: read-president president <- read_csv("data/us_presidents.csv", show_col_types = FALSE) slice_sample(president, n = 10) |> gt() |> tab_header(title = "president") ``` She realized that her data only had the start/end dates for each president and she wanted a data set that filled in the missing years and political party for the president in that time period. After much googling and reading Stack Overflow, found two functions she did not know about called `complete()` and `fill()` to fill in the missing years and party affiliations ```{r} #| label: filling-in-year politics_year <- president |> mutate(start = lubridate::mdy(start), #formats date correctly start_year = lubridate::year(start)) |> #pulls out year filter(start_year > 1940) |> #removes data before 1940 since there was no UN select(start_year, party) |> #pulls out just the variables of interest complete(start_year = seq(min(start_year), 2020, by = 1)) |> #fill in missing years fill(party) #fill in missing party affiliations for years slice_tail(politics_year, n = 10) |> gt() |> tab_header(title = "Year by Presidential Party") ``` Next, your instructor took the `yes_votes` data and filtered out just the United States data and then joined by year to add the party affiliation of the president for each year of UN votes. To create the visualization with the smoothed model, but color by party affiliation, she had to add a new column called `predict` that fit the model first instead of using `geom_smooth()`. ```{r} #| label: full-clean-president-graph yes_votes |> filter(country == "United States") |> #we want to focus on the US left_join(politics_year, by = c("year" = "start_year")) |> #adding in president political affiliation data group_by(issue) |> #we want to calculate predicted yes by issue type mutate(predict = predict(loess(prop_yes ~ year))) |> #creates predictions using a loess model for yes vs year ungroup() |> #ungroups the group_by line so it doesn't mess with other calculations ggplot(mapping = aes(x = year, color = party, group = 1)) + #sets universal aes to observe changes over time geom_point(aes(y = prop_yes), alpha = 0.4) + #plots points for proportion yes but transparent geom_line(aes(y = predict)) + #plots the predicted yeses for votes facet_wrap(~issue) + #creates a plot for each issue scale_y_continuous(labels = scales::percent) + #scales y-axis as percent values labs( title = "Percentage of United States 'Yes' votes in the UN General Assembly", subtitle = "1946 to 2019", y = "% Yes", x = "Year", color = "Presidential Political Party" ) + #adds labels theme_bw() + #changes appearance scale_color_manual(values = c("blue", "red"), breaks = c("Democratic", "Republican")) + #sets specific colors for each party theme(legend.position = "bottom") #moves legend to bottom to create space ```