Chunk setup
Set your working directory. Due to encoding issues, we need to set
locale to “Norwegian_Bokmål_Norway”.
knitr::opts_chunk$set(warning=FALSE, message=FALSE, dpi = 300, dev='png', fig.path = "figures_fr/")
Sys.setlocale(locale = 'Norwegian Bokmål_Norway')
## Warning in Sys.setlocale(locale = "Norwegian Bokmål_Norway"): using locale code
## page other than 65001 ("UTF-8") may cause problems
## [1] "LC_COLLATE=Norwegian Bokmål_Norway.utf8;LC_CTYPE=Norwegian Bokmål_Norway.1252;LC_MONETARY=Norwegian Bokmål_Norway.utf8;LC_NUMERIC=C;LC_TIME=Norwegian Bokmål_Norway.utf8"
Load packages used in the analysis
library(rvest) # for scraping html-files
library(tidyverse) # ggplot and datamangement
library(haven) # for import of data
library(readxl) # reading xl files
library(ggrepel) # for labelling plots
library(marginaleffects) # marginal effects
library(gt) # tables
library(gtExtras) # tables
library(modelsummary) # regression tables
library(broom) # tidy regression results
library(gghalves) # for plotting the boxplots with points
library(ggbeeswarm) # for plotting the boxplots with points
library(scales) # for plotting
1. Description
The code in this document generate the data and results reported in
Differential engagement with constitutional and international
fundamental rights on the Supreme Court of Norway. The first part
of this document creates the dataset and the second part reports the
analysis, as well as complementary results.
The final data is created from two different sources. The following
lists the most important variables that we use from each dataset and
that we use for further coding purposes.
2. The Rights dataset
We created a dataset Rights that links rights
articles (e.g. EHCR art. 6) to its substantive content (e.g. right
to a fair trial) and then categorized each article in six
non-exclusive categories based on the Comparative Constitutions Project
(CCP) to measure the substantive content of the fundamental right. The
rights that are included in the data are the fundamental rights that
Norway have ratified through the Human Rights Act, with later
amendments (see https://lovdata.no/dokument/NLE/lov/1999-05-21-30)
The six rights categories are:
civil and political rights
economic rights
equality, gender, and minority rights
legal procedural rights
physical integrity rights
social rights
The CCP already coded the Norwegian Constitution. We coded each
article in ECHR using a dictionary-based approach and we hand-coded the
ICCPR, ICESCR, CRC and CEDAW.
In total, the dataset consist of 190 rights articles. The data is
created by running the rights.R-script below. In the analysis we use
this data to examine how much judges use different FR laws under
different categories of rights.
3. Data collection: creating law citation data from unstrutured text
data from LovdataPro
Using a LovdataPro account we identified fundamental rights
cases following this procedure. We used a set of selection criteria to
create a subset of fundamental rights cases:
- rettsområde (area of law) == “Human Rights” or “Constitutional
law”,
- court == the Supreme Court,
- Kategori (category) == “Avdeling” (also include grand chamber or
plenary cases)
- Periode (period) == 2008-2022
This results in 233 decisions with references to fundamental rights
laws from 2008-2022.
We download all cases, including metadata, as HTML-files and extract
information from decisions using the rvest package
(Wickham 2023) in the statistical software R.
3.1 Get html files
From each decision we extract metadata on the type of decision
(judgement/interlocutory judgment), date of the decision, case keywords,
case summary, case proceedings, case parties, judge panel, and last
updated.
The table function extracts metadata from Lovdatata’s
summary field
table <- function(x) {
x %>%
html_element("table") %>%
html_table() %>%
pivot_wider(names_from = X1, values_from = X2)
}
Apply table function to cases:
metadata <- lapply(cases, table)
Turn metadata into a dataframe object:
df_summary <- bind_rows(metadata)
df_summary have nine columns. The nine columns are (english
translation): short description
- Instans (Authority): name of court - type of decision
- Dato (Date): decision date
- Publisert (Published): casenumber
- Stikkord (Keywords): n words describing area(s) of law, facts and
legal question
- Sammendrag (Summary): decision summary
- Saksgang (Proceedings): case history
- Parter (Parties): appellant v. respondent
- Forfatter (Author): judge names
- Sist oppdatert (Last update): last time decision text was
updated
We create different casenumbers variables for later merging purposes:
df_summary <- df_summary %>%
mutate(Publisert = str_replace_all(Publisert, "\\–", "\\-"))
df_summary <- df_summary %>%
mutate(hr_number = str_extract_all(Publisert, "\\w+\\-\\d+\\-\\d+\\-\\w"),
rt_number = str_extract_all(Publisert, "Rt\\-\\d+\\-\\d+"),
published = ifelse(rt_number == "character(0)", as.character(hr_number), as.character(rt_number)))
We use information in df_summary to create case-level variables of
interest
- type_of_decision: two different types and three
cases that we will remove. The relevant information is contained in the
variable Instans.
Cases in the remove from sample category are the Supreme Court’s are
one-judge decisions on postponement of processing of appeals
(HR-2009-656-F and HR-2009-746-F). One case is is SCONs “betenkning” to
the Norwegian parliament (HR-2021-655-P).
judgment <- c("Dom|Dom og kjennelse|Dom og orskurd") # cases that are both are classified as judgment
interlocutory_judgment <- ("Kjennelse|Orskurd")
remove_from_sample <- ("Beslutning|Betenkning")
df_summary <- df_summary %>%
mutate(
type_of_decision = case_when(
grepl(judgment, Instans) ~ "Decision",
grepl(interlocutory_judgment, Instans) ~ "Order",
grepl(remove_from_sample, Instans) ~ "remove_from_sample"
)
)
- dissent case: whether case is decided with
dissenting opinions or not. This is information is available in the
string variable Forfatter (Author), which lists judge names and
dissenters.
dissent <- c("Dissens|dissens|Særmerknader") # this id all types of disagreement
df_summary <- df_summary %>%
mutate(dissent_case = ifelse(grepl(dissent, Forfatter), "Dissents", "Consensus"))
- type of panel: five-justice panel, grand chamber or
plenary. In the string variable Publisert “A” is short for “Avdeling”
(regular cases decided in grand chamber), “S” is short for “Storkammer”
(case decided in strengthened panel with 11 judges or less), and “P” is
short for Plenum (cases decided by all judges on the court in
plenary).
five_judge <- c("A")
grand_chamber <- ("S")
plenary <- ("P")
df_summary <- df_summary %>%
mutate(
type_of_panel = case_when(
grepl(five_judge, Publisert) ~ "Five-judge panel",
grepl(grand_chamber, Publisert) ~ "Grand chamber",
grepl(plenary, Publisert) ~ "Plenary"
)
)
- Civil or criminal law: In case keywords, criminal
law cases are identified by the word “straff” (criminal), civil law
cases are identified as not including this word in case keyword.
df_summary <- df_summary %>%
mutate(type_of_law = ifelse(str_detect(tolower(Stikkord), "straff"), "Criminal law", "Civil law"))
There are four plenary cases that concern the impartiality of the
Supreme Court justices participating in the decision. We want to remove
these from the analysis later on. Here we simply identify them.
df_summary <- df_summary %>%
mutate(plenary_impartiality = case_when(type_of_panel == "Plenary" & str_detect(tolower(Stikkord), "habilitet") ~ "Yes"))
3.1.1 Get law citation links
From written decisions we extract text, hypertext, hyperlinks, as
well as paragraph number, and then link these data together in a matrix
where each row is a law citation belonging to a paragraph
of text belonging to a decision. Paragraphs that do not
include hyperlinks to laws are empty cells.
This code extracts casenumber, hypertext and hyperlinks from
decisions and create a dataframe with three variables
links <- cases %>%
map_df(~{
case <- .x %>% html_nodes("h1") %>% html_text()
name <- .x %>% html_nodes("a") %>% html_text()
links <- .x %>% html_nodes("a") %>% html_attr("href")
tibble(case, name, links)
}
)
3.1.2 Get paragraphs of text
This code extracts paragraphs of text and casenumber and create a
dataframe with three variables
par <- cases %>%
map_df(~{
case <- .x %>% html_nodes("h1") %>% html_text()
nodes <-
xml2::xml_find_all(
.x,
".//p[preceding-sibling::div[@align='center']] | .//table[preceding-sibling::div[@align='center']] | .//a[preceding-sibling::span]|.//a[following-sibling::span]"
)
text <- rvest::html_text(nodes)
text <- gsub("_", "", text)
ref_link <- rvest::html_attr(nodes, "href")
tibble(case, text, ref_link)
}
)
This code generate id variable that identify paragraphs within case.
Data in the variable text and that have missing values in par_paragraph
is from the summary of the decision and is not part of the decision
itself. The variable link_name extract the hypertext associated with the
hyperlinks.
par <- par %>%
mutate(par_paragraph = str_extract(text, "(^\\(\\d+)\\)"),
par_paragraph = str_replace_all(par_paragraph, "[(]|[)]", ""),
par_paragraph = as.numeric(as.character(par_paragraph)))
par <- par %>%
mutate(link_name = ifelse(!str_detect(text, "(^\\(\\d+)\\)"), text, NA))
par <- par %>%
group_by(case) %>%
fill(par_paragraph)
Paragraphs with text:
par_text <- par %>%
dplyr::select(-c(ref_link, link_name)) %>% # remove ref_link
filter(str_detect(text, "(^\\(\\d+)\\)")) %>%
mutate(par_text = gsub("^\\(\\d+\\)", "", text))
Paragraphs with law citations:
par_references <- par %>%
dplyr::select(-text) %>%
filter(!is.na(par_paragraph)) %>%
filter(!is.na(ref_link))
Join par_text and par_references in a new data frame where
ref_link is the unit of observation and belongs to a paragraph
of text (par_paragraph and par_text) which belongs to a case. Paragraphs
without ref_link are NA in ref_link.
text_and_links <- par_text %>%
full_join(par_references, by = c("case", "par_paragraph")) %>%
dplyr::select(-text)
remove(par_references, par_text, par, cases) # remove three df's that we do not need anymore
Make Publisert variable to merge with df_summary:
publisert <- unique(df_summary$Publisert)
publisert <- paste(publisert, collapse = "|")
text_and_links <- text_and_links %>%
mutate(Publisert = str_extract_all(case, publisert)) %>%
unnest(cols = Publisert)
3.1.3 Identify judges to make par_judge variable
We use information in the string-variable par_text to identify the
author of the paragraph text.
# text_and_links <- text_and_links %>%
# mutate(par_text = gsub("Dommer :", "Dommer", par_text))
Text patterns to identify judges:
text_and_links <- text_and_links %>%
mutate(par_judge = ifelse(
grepl(
"^ *Domm[ea]r[ne]* .*:.*$|^ *Justit[ui]arius .*:.*$|^ *Kst\\. domm[ea]r .*:.*$",
par_text
),
gsub(
"^ *(Domm[ea]r[ne]*|Justit[ui]arius|Kst\\. domm[ea]r) (.*?):.*$",
"\\2",
par_text
),
NA
),
par_judge = strsplit(gsub("\\.", "", par_judge), " og |, "))
3.1.5 Fill judge variable
Now, we fill the par_judge variable. Remember that we want to
separate the judgement part from the judge author part.
text_and_links <- text_and_links %>%
mutate(
par_judge = as.character(par_judge)) %>%
fill(par_judge)
Create judge order variable for coding of dissenting opinions
judge_order <- text_and_links %>%
group_by(case) %>%
distinct(par_judge) %>%
mutate(
judge_order = seq_along(par_judge))
text_and_links <- text_and_links %>%
inner_join(judge_order, by = c("case", "par_judge"))
3.1.4 Make par_section variable that id subsections of decision
We create a variable that identify subsections in the decision. The
structure of each decision looks like this:
- Syllabus
- Main legal question
- Proceedings
- The plaintiff’s claims
- Legal arguments
- Legal claims
- The respondent’s claims
- Legal arguments
- Legal claims
- Main opinion
- Majority opinion
- Minority opinion(s)
- Votes
- Judgement
Syllabus refers to the introductory parts of decisions. It normally
consists of 1) a general statement as to what is the main legal question
in the case, 2) proceedings, the legal history of the case, 3) the
plaintiffs’ claim, and 4) the respondents’ claim. From the syllabus
section our main interest is to identify the parties fundamental rights
arguments.
The following text patterns are gathered from Bjørnebekk and
Johannesson (2016) and identify the start of each section. The syllabus
section is identified by starting at par_paragraph 1.
Main opinion text patterns:
main_opinion <-
c("^ *Eg går så over til å gjere greie for mitt syn på saka|^ *Eg går no over til å gjere greie for mitt syn på saka|^ *Eg går no over til å gi uttrykk for mitt syn på habilitetsspørsmåla|^ *Jeg er kommet til|^ *Jeg har kommet til|^ *Min vurdering av saken|^ *Jeg ser langt på vei saken på samme måte som de tidligere retter|^ *Jeg starter med å se|^ *Jeg finner at|^ *Jeg bemerker at saken|^ *Mitt syn på saken|^ *Eg er komen til|^ *Mitt syn på saka|^ *Egne bemerkninger|^ *Jeg ser først på|^ *Jeg ser slik på saken:|^ *Jeg er – under atskillig tvil – kommet til at anken fører frem|^ *Jeg er når det gjelder utmålingen |^ *Jeg er blitt stående ved at anken bør tas til følge|^ *Jeg går nå over til å gi uttrykk for mitt syn på saken|^ *Jeg finner det naturlig å behandle anken over saksbehandlingen først|^ *Høyesteretts syn på saken")
Judgement text patterns:
judgement <- c("^ *Etter stemmegivningen avsa Høyesterett denne|^ *Etter røystinga sa Høgsterett slik|^ *Kjennelsen er enstemmig")
Votes text patterns:
votes_1 <- c("^ *Eg røystar etter dette|[Jj]eg stemmer for|^dom:$|^ *Jeg stemmer etter dette|^ *Etter dette stemmer jeg for|^Da jeg er i mindretall, former jeg ingen konklusjon")
Text patterns that identify judges who join opinions, i.e., who do
not write their own opinions
judge_join_opinion <-
c(
" *\\:\\sLikeså\\.| *:\\sDet same\\.| *:\\sJeg er i det vesentlige og i resultatet enig med \\w+\\, dommer \\w+\\.| *:\\sJeg er i det vesentlige og i resultatet enig med \\w+\\.| *:\\sEg er i det hovudsaklege og i resultatet einig med \\w+\\, dommar \\w+\\.| *:\\sEg er i det hovudsaklege og i resultatet einig med \\w+\\.| *\\:\\sSom dommer \\w+\\."
)
text_and_links <- text_and_links %>%
mutate(join_opinion = case_when(grepl(judge_join_opinion, par_text) ~ "Join opinion"))
Create the par_section variable:
text_and_links <- text_and_links %>%
mutate(
par_section = case_when(
judge_order == 1 & grepl(main_opinion, par_text) ~ "Main opinion",
grepl(judgement, par_text) ~ "Judgement",
grepl(votes_1, par_text) ~ "Vote",
grepl(judge_join_opinion, par_text) ~ "Join opinion",
par_paragraph == 1 ~ "Syllabus",
judge_order != 1 & !grepl(judge_join_opinion, par_text) ~ "Dissenting opinion"
)
)
Fill par_section variable:
text_and_links <- text_and_links %>%
fill(par_section)
Recode dissenting opinions with more than one dissent
text_and_links <- text_and_links %>%
group_by(case, par_section) %>%
mutate(dissent_order = match(par_judge, unique(par_judge))) %>%
ungroup()
text_and_links <- text_and_links %>%
mutate(par_section = ifelse(par_section == "Dissenting opinion", paste(par_section, dissent_order, sep = " "), par_section))
To code characteristics of the parties claims we combined regular
expressions and manual coding. We collected data manually on the last
paragraph number in the syllabus, the first and last paragraph number in
the plaintiffs’ claims. From these three numbers we identified the
syllabus sections of all decisions and then coded which paragraphs that
are plaintiff claims and respondent claims.
Read syllabus data:
Make case number for merging
syllabus <- syllabus %>%
mutate(number = str_replace_all(case, "[^\\w\\d]", ""))
Do the same in the text_and_link data
text_and_links <- text_and_links %>%
mutate(number = str_replace_all(case, "[^\\w\\d]", ""))
Merge syllabus data with text_link_summary data.
text_and_links <- text_and_links %>%
left_join(syllabus, by = "number")
Three cases that do not merge are cases that we want to remove from
the sample. They are:
Norges Høyesterett - Beslutning. HR-2009-656-F
Norges Høyesterett - Beslutning. HR-2009-746-F
Norges Høyesterett - Betenkning HR-2021-655-P
Make syllabus section variables:
text_and_links <- text_and_links %>%
mutate(
syllabus_sections = case_when(
par_paragraph >= plantiff_start &
par_paragraph <= plaintiff_ends ~ "Plaintiff",
par_paragraph > plaintiff_ends &
par_paragraph <= syllabus_ends ~ "Respondent",
par_paragraph == 1 | par_paragraph == 2 & grepl("Saka gjeld|Saken gjelder", par_text) ~ "Main question"
)
)
Make detailed case section variable
text_and_links <- text_and_links %>%
mutate(
syllabus_sections = ifelse(
is.na(syllabus_sections) &
par_paragraph < syllabus_ends,
"Case history",
syllabus_sections
),
section_detailed = ifelse(is.na(syllabus_sections), join_opinion, syllabus_sections),
section_detailed2 = ifelse(is.na(section_detailed), par_section, section_detailed),
section_detailed2 = ifelse(par_section == "Judgement", par_section, section_detailed2)
)
Remove three variables we do not need anymore
text_and_links <- text_and_links %>%
dplyr::select(-c(join_opinion, section_detailed, syllabus_sections))
The section_detailed2 variable is wrong for 44 cases
check <- text_and_links %>% filter(section_detailed2 == "Syllabus" & par_paragraph>syllabus_ends)
length(unique(check$case.x))
## [1] 44
Recode section_detailed2 to “Main opinion” if category equal syllabus
and par_paragraph (paragraph number) is bigger than the number in
syllabus_ends.
text_and_links <- text_and_links %>%
mutate(section_detailed2 = ifelse(section_detailed2 == "Syllabus" & par_paragraph>syllabus_ends, "Main opinion", section_detailed2))
Running the test again, we find 0 errors
check <- text_and_links %>% filter(section_detailed2 == "Syllabus" & par_paragraph>syllabus_ends)
length(unique(check$case.x))
## [1] 0
We change four judgenames for later merging purposes “Schei_C” ==
Schei, Endresen, C == Endresen, Stabel, I == Stabel, Øie_C == Øie if
date > feb 2016.
text_and_links <- text_and_links %>%
mutate(jfamname = case_when(par_judge == "Schei" ~ "Schei_C",
par_judge == "Endresen" ~ "Endresen, C",
par_judge == "Stabel" ~ "Stabel, I",
par_judge == "Øie" & str_detect(par_text, "Justitiarius") ~ "Øie_C",
TRUE ~ as.character(par_judge)))
3.1.5 Merge text and links with df_summary
text_links_summary <- text_and_links %>%
left_join(df_summary, by = "Publisert")
3.2 Work with links
We use information in the string variable ref_link to identify law
citations and to categorize them at different levels of abstraction.
We use the ref_link to extract law citations. Table below shows 10
different FR law sources and their corresponding matching pattern.
| The Constitution of the Kingdom of Norway |
“1814” |
| International Covenant on Economic, Social and Cultural Rights |
“1999-05-21-30/oskn” |
| International Covenant on Civil and Political Rights |
“1999-05-21-30/spn” |
| Convention on the Elimination of All Forms of Discrimination against
Women |
“1999-05-21-30/kdkn” |
| Convention on the Rights of the Child |
“1999-05-21-30/bkn” |
| European Convention on Human Rights |
“1999-05-21-30/emkn|1999-05-21-30/emke” |
| ECtHR Caselaw |
“emd\-\d+\d+-” |
| UN General Comment The Committee on the Rights of Persons with
Disabilities |
“gc\-\d+\d+\-\d+\-crpd” |
| UN General Comment The Committee on Economic, Social and Cultural
Rights |
“gc\-\d+\d+\-\d+\-cescr” |
| UN General Comment The Committee on Civil and Political Rights |
“gc\-\d+\d+\-\d+\-ccpr” |
| UN General Comment The Committee on the Rights of the Child |
“gc\-\d+\d+\-\d+\-crc” |
| CAT Caselaw |
“cat\-\d+\d+” |
| ICCPR Caselaw |
“ccpr\-\d+\d+” |
Create a variable with the official names of human rights
conventions:
text_links_summary <- text_links_summary %>%
mutate(law_name = ifelse(
grepl("1814", ref_link),
"The Constitution of the Kingdom of Norway",
ifelse(
grepl("1999-05-21-30/oskn", ref_link),
"International Covenant on Economic, Social and Cultural Rights",
ifelse(
grepl("1999-05-21-30/spn", ref_link),
"International Covenant on Civil and Political Rights",
ifelse(
grepl("1999-05-21-30/oskn", ref_link),
"International Covenant on Economic, Social and Cultural Rights",
ifelse(
grepl("1999-05-21-30/kdkn", ref_link),
"Convention on the Elimination of All Forms of Discrimination against Women",
ifelse(
grepl("1999-05-21-30/bkn", ref_link),
"Convention on the Rights of the Child",
ifelse(
grepl("1999-05-21-30/emkn|1999-05-21-30/emke", ref_link),
"European Convention on Human Rights",
ifelse(
grepl("emd\\-\\d+\\d+-", ref_link),
"ECtHR Caselaw",
ifelse(
grepl("gc\\-\\d+\\d+\\-\\d+\\-crpd", ref_link),
"UN General Comment The Committee on the Rights of Persons with Disabilities",
ifelse(
grepl("gc\\-\\d+\\d+\\-\\d+\\-cescr", ref_link),
"UN General Comment The Committee on Economic, Social and Cultural Rights",
ifelse(
grepl("gc\\-\\d+\\d+\\-\\d+\\-ccpr", ref_link),
"UN General Comment The Committee on Civil and Political Rights",
ifelse(
grepl("gc\\-\\d+\\d+\\-\\d+\\-crc", ref_link),
"UN General Comment The Committee on the Rights of the Child",
ifelse(
grepl("cat\\-\\d+\\d+", ref_link),
"CAT Caselaw",
ifelse(
grepl("ccpr\\-\\d+\\d+", ref_link),
"ICCPR Caselaw",
ifelse(grepl("ccrr\\-\\d+\\d+", ref_link),
"ICCPR Caselaw",
NA)
)
)
)
)
)
)
)
)
)
)
)
)
)
))
Number of references to different FR laws:
text_links_summary %>%
count(law_name) %>%
gt()
| law_name |
n |
| Convention on the Elimination of All Forms of Discrimination against Women |
5 |
| Convention on the Rights of the Child |
293 |
| ECtHR Caselaw |
954 |
| European Convention on Human Rights |
2562 |
| ICCPR Caselaw |
64 |
| International Covenant on Civil and Political Rights |
373 |
| International Covenant on Economic, Social and Cultural Rights |
7 |
| The Constitution of the Kingdom of Norway |
1386 |
| UN General Comment The Committee on Civil and Political Rights |
14 |
| UN General Comment The Committee on Economic, Social and Cultural Rights |
2 |
| UN General Comment The Committee on the Rights of Persons with Disabilities |
1 |
| UN General Comment The Committee on the Rights of the Child |
37 |
| NA |
23221 |
We identify citations to all other types of legal sources, by
extracting Lovdata’s legal source classification category from the
hyperlink. Table below shows 10 different law sources and their
corresponding matching pattern.
| Statutory law |
gsub(“https\:\/\/lovdata\.no\/pro\#reference\/”, ““) |
| Case law |
|
| Preparatory works |
|
| Regulations |
|
| Circulars |
|
| Academic literature |
|
| Treaty |
|
| EU |
|
| http: |
|
| https: |
|
Citation type (law, preporatory works, regulations, case law
etc):
text_links_summary <- text_links_summary %>%
mutate(citation_type = gsub("https\\:\\/\\/lovdata\\.no\\/pro\\#reference\\/", "", ref_link),
citation_category = gsub("/.*", "", citation_type),
law_name = ifelse(is.na(law_name), citation_category, law_name))
Number of references to different types of laws:
# summarytools::freq(text_links_summary$law_name[which(!is.na(text_links_summary$law_name ))])
text_links_summary %>%
filter(!is.na(law_name)) %>%
group_by(law_name) %>%
count() %>%
ggplot(aes(y=n, x = reorder(law_name, -n))) +
geom_col() +
labs(x = "") +
coord_flip() +
theme_minimal()
3.2.1 Generate article variable
First, create raw_reference from ref_link:
text_links_summary<- text_links_summary %>%
mutate(raw_reference = str_replace_all(ref_link, "https\\:\\/\\/lovdata\\.no\\/pro\\#reference\\/\\w+\\/", ""))
Separate citations to the constitution from citations to other
Norwegian laws:
text_links_summary<- text_links_summary %>%
mutate(raw_reference = str_replace_all(raw_reference, "^1814-05-17/§", "1814-05-17/grl"))
text_links_summary<- text_links_summary %>%
separate(raw_reference, c("date", "b", "c", "d" ), sep = "\\/", remove = F)
text_links_summary <- text_links_summary %>%
unite("article", b:d, sep = " ")
text_links_summary<- text_links_summary %>%
mutate(article = str_replace_all(article, "NA", ""))
string trim end of string:
text_links_summary<- text_links_summary %>%
mutate(article = str_trim(article, side = "right"))
Replace “” with NA:
text_links_summary <- text_links_summary %>%
mutate(article = gsub("^$", NA, trimws(article)))
Add case law to article, problem with case law-paragraph
citations:
text_links_summary <- text_links_summary %>%
mutate(article = ifelse(is.na(article), raw_reference, article),
article = ifelse(citation_category == "avgjorelse"| citation_category == "forabeid", raw_reference, article))
Replace “emke” with “emkn”:
text_links_summary<- text_links_summary %>%
mutate(article = str_replace_all(article, "emke", "emkn"))
Make case date variable:
text_links_summary<- text_links_summary %>%
mutate(case_date = as.Date(Dato))
Identify references with rights content that change over time
text_links_summary <- text_links_summary %>%
mutate(article = ifelse(
article == "grl2" & case_date < "2012-05-21",
"rpre2012_grl2",
ifelse(
article == "grl16" & case_date >= "2012-05-21",
"rpost2012_grl16_r",
ifelse(
article == "grl92" & case_date >= "2014-05-14",
"rpost2014_grl92",
ifelse(
article == "grl93" & case_date >= "2014-05-14",
"rpost2014_grl93",
ifelse(
article == "grl94" & case_date >= "2014-05-14",
"rpost2014_grl94",
ifelse(
article == "grl95" & case_date >= "2014-05-14",
"rpost2014_grl95",
ifelse(
article == "grl96" & case_date < "2014-05-14",
"rpre2014_grl96",
ifelse(
article == "grl96" & case_date >= "2014-05-14",
"rpost2014_grl96",
ifelse(
article == "grl97",
"r_grl97",
ifelse(
article == "grl98" & case_date >= "2014-05-14",
"rpost2014_grl98",
ifelse(
article == "grl99" & case_date < "2014-05-14",
"rpre2014_grl99",
ifelse(
article == "grl100",
"r_grl100",
ifelse(
article == "grl101" & case_date < "2014-05-14",
"rpre2014_grl101",
ifelse(
article == "grl101" & case_date >= "2014-05-14",
"rpost2014_grl101",
ifelse(
article == "grl102" & case_date < "2014-05-14",
"rpre2014_grl102",
ifelse(
article == "grl102" & case_date >= "2014-05-14",
"rpost2014_grl102",
ifelse(
article == "grl103" & case_date < "2014-05-14",
"rpre2014_grl103",
ifelse(
article == "grl104" & case_date >= "2014-05-14",
"rpost2014_grl104",
ifelse(
article == "grl105",
"r_grl105",
ifelse(
article == "grl106" & case_date >= "2014-05-14",
"rpost2014_grl106",
ifelse(
article == "grl108" & case_date >= "2014-05-14",
"rpost2014_grl108",
ifelse(
article == "grl109" & case_date >= "2014-05-14",
"rpost2014_grl109",
ifelse(
article == "grl110" & case_date < "2014-05-14",
"rpre2014_grl110",
ifelse(
article == "grl110" & case_date >= "2014-05-14",
"rpost2014_grl110",
ifelse(
article == "grl110a" & case_date < "2014-05-14",
"rpre2014_grl110a",
ifelse(
article == "grl110b" & case_date < "2014-05-14",
"rpre2014_grl110b",
ifelse(
article == "grl110c" & case_date < "2014-05-14",
"rpre2014_grl110c",
ifelse(
article == "grl112" & case_date >= "2014-05-14",
"rpost2014_grl112",
ifelse(
article == "grl113" &
case_date >= "2014-05-14",
"rpost2014_grl113",
article
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
))
Get general references to the 1) human rights act and 2) the
constitution:
text_links_summary<- text_links_summary %>%
mutate(article = ifelse(date == "1999-05-21-30" & law_name == "lov", raw_reference,
ifelse(ref_link == "1814-05-17", "constitution", article)))
Human rights act:
text_links_summary <- text_links_summary %>%
mutate(law_name = ifelse(law_name == "lov" & date == "1999-05-21-30", "Human Rights Act", law_name))
Merge with rights-dataset
text_links_summary <- text_links_summary %>%
full_join(rights, by = "article")
Remove rights that are never cited by scon
text_links_summary <- text_links_summary %>%
filter(!is.na(case.x))
Counting ECHR-references shows that the right to privacy
(art.8) and the right to a fair trial (art. 6) is cited much
more frequently than other ECHR-articles
The most cited section in the constitution is the ban on retroactive
laws in (section 97). Many of most cited constitutional rights are
rights that were added in the constitutional amendment in 2014.
The most cited rights in ICCP are article 14 on right to a fair
trial, article 27 on the right to culture and article 19 on freedom of
expression.
The most cited rights in CRC is article 3 on the best interests of
the child, article 12 on the right to express views freely in all
matters affecting the child, and article 9 on separation.
3.3 Create a FR law dictionary that pick ups law citiations that are
left out
By eyeballing the citation and text data, we can see that not all law
citations in the decisions are embedded as hyperlinks in the HTML
files.
More systematic examination reveal quite large discrepancies between
actual citations and hyperlinks.
To illustrate, from the extracted hyperlinks we identify 2562
references to ECHR
text_links_summary %>%
count(law_name) %>%
filter(law_name == "European Convention on Human Rights")
## # A tibble: 1 x 2
## law_name n
## <chr> <int>
## 1 European Convention on Human Rights 2562
Yet searching for the text pattern “ECHR article \d+” in the same
data recovered 4088 mentions.
sum(str_count(text_links_summary$par_text,"EMK artikkel| EMK art."))
## [1] 4088
Similarly, from extracted hyperlinks we discovered 1386 citations to
the Norwegian constitution
text_links_summary %>%
count(law_name) %>%
filter(law_name == "The Constitution of the Kingdom of Norway")
## # A tibble: 1 x 2
## law_name n
## <chr> <int>
## 1 The Constitution of the Kingdom of Norway 1386
Yet searching for the text pattern “Grunnloven § \d+|Grunnlova § \d+”
in the same data we recover 3416 mentions.
sum(str_count(text_links_summary$par_text,"Grunnloven § \\d+|Grunnlova § \\d+"))
## [1] 3416
For UN convention om the rights of the child we find 293
citations
text_links_summary %>%
count(law_name) %>%
filter(law_name == "Convention on the Rights of the Child")
## # A tibble: 1 x 2
## law_name n
## <chr> <int>
## 1 Convention on the Rights of the Child 293
and 404 mentions
sum(str_count(text_links_summary$par_text,"barnekonvensjonen artikkel| barnekonvensjonen art."))
## [1] 404
The simplest explanation for the difference between hyperlinks and
law mentions is that Lovdata forget to embed law mentions as
hypertext.
When judges write opinions they reference the same laws in different
ways (e.g. ECHR art. 8 and ECHR article 8). The many different ways that
judges cite similar laws are collected in the hypertext in the
links-dataframe.
Assuming that Lovdata’s errors (forgetting to embed law mentions as
hypertext) concern laws they at one point in time have embedded as
hypertext, we can use the hypertext data to fix Lovdata’s error.
We use the hypertext to generate a dictionary of different ways of
citing fundamental rights sources. Some hypertext, however, do not
identify which type of FR law that is cited (e.g., the text is simply
“article 3”, which can refer to both CRC and EHCR). We remove these
entries from the dictionary because they cast a too wide net. Before we
remove these entries we need to check whether there are any paragraphs
that include these specific string patterns and that is not already
cited.
# The dictionary
# BKN
bkn <- unique(links$name[which(str_detect(links$links, "bkn"))])
bkn <- tolower(bkn)
bkn <- str_remove_all(bkn, "\\(")
bkn <- str_remove_all(bkn, "\\)")
bkn <- unique(bkn)
bkn <- bkn[which(str_detect(bkn, "^b"))]
#add manual entries of mentions
bkn <- append(bkn, "barnekomit*")
# SPN
spn <- unique(links$name[which(str_detect(links$links, "spn"))])
spn <- tolower(spn)
spn <- str_remove_all(spn, "\\(")
spn <- str_remove_all(spn, "\\)")
spn <- unique(spn)
spn <- spn[which(str_detect(spn, "^s|^k"))]
spn <- str_replace(spn, "^sp$", "\\bsp\\b")
# ECHR
echr <- unique(links$name[which(str_detect(links$links, "emkn"))])
echr <- tolower(echr)
echr <- str_remove_all(echr, "\\(")
echr <- str_remove_all(echr, "\\)")
echr <- unique(echr)
echr <- str_trim(echr)
echr <- echr[which(str_detect(echr, "^e|^m|^p|^d"))]
# CONSTITUTION
grl <- unique(links$name[which(str_detect(links$links, "1814-05-17"))])
grl <- tolower(grl)
grl <- str_remove_all(grl, "\\(")
grl <- str_remove_all(grl, "\\)")
grl <- unique(grl)
grl <- grl[which(str_detect(grl, "^g"))]
Before we apply the dictionary we make all text lower-case and remove
parenthesis.
text_links_summary <- text_links_summary %>%
mutate(text = tolower(par_text))
text_links_summary <- text_links_summary %>%
mutate(text = str_remove_all(text, "\\("))
text_links_summary <- text_links_summary %>%
mutate(text = str_remove_all(text, "\\)"))
First, we code paragraphs based on hyperlinks. We create one column
for each type of FR law source (e.g. ECHR_in_paragraph), which identify
whether that FR law source is cited in the paragraph.
paragraphs_text <-
text_links_summary %>%
group_by(case.x, par_paragraph) %>%
mutate(
echr_in_paragraph = ifelse(any(
str_detect(law_name, "European Convention on Human Rights")
), "European Convention on Human Rights", NA),
ecthr_in_paragraph = ifelse(any(str_detect(
law_name, "ECtHR Caselaw"
)), "ECtHR Caselaw", NA),
constitution_in_paragraph = ifelse(any(
str_detect(law_name, "The Constitution of the Kingdom of Norway")
), "The Constitution of the Kingdom of Norway", NA),
crc_in_paragraph = ifelse(any(
str_detect(law_name, "Convention on the Rights of the Child")
), "Convention on the Rights of the Child", NA),
iccpr_in_paragraph = ifelse(
any(
str_detect(law_name, "International Covenant on Civil and Political Rights")
),
"International Covenant on Civil and Political Rights",
NA
),
icescr_in_paragraph = ifelse(
any(
str_detect(
law_name,
"International Covenant on Economic, Social and Cultural Rights"
)
),
"International Covenant on Economic, Social and Cultural Rights",
NA
),
un_gc_crc_in_paragraph = ifelse(
any(
str_detect(
law_name,
"UN General Comment The Committee on the Rights of the Child"
)
),
"UN General Comment The Committee on the Rights of the Child",
NA
),
un_gc_ccpr_in_paragraph = ifelse(
any(
str_detect(
law_name,
"UN General Comment The Committee on Civil and Political Rights"
)
),
"UN General Comment The Committee on Civil and Political Rights",
NA
),
un_gc_cescr_in_paragraph = ifelse(
any(
str_detect(
law_name,
"UN General Comment The Committee on Economic, Social and Cultural Rights"
)
),
"UN General Comment The Committee on Economic, Social and Cultural Rights",
NA
),
un_gc_crpd_in_paragraph = ifelse(
any(
str_detect(
law_name,
"UN General Comment The Committee on the Rights of Persons with Disabilities"
)
),
"UN General Comment The Committee on the Rights of Persons with Disabilities",
NA
),
iccpr_caselaw_in_paragraph = ifelse(any(str_detect(
law_name, "ICCPR Caselaw"
)), "ICCPR Caselaw", NA),
cedaw_in_paragraph = ifelse(
any(
str_detect(
law_name,
"Convention on the Elimination of All Forms of Discrimination against Women"
)
),
"Convention on the Elimination of All Forms of Discrimination against Women",
NA
)
)
Then we code paragraphs based on hypertext and codebook
paragraphs_text <- paragraphs_text %>%
distinct(case.x, par_paragraph, .keep_all = T) %>%
mutate(
echr_in_paragraph_dict = ifelse(any(str_detect(text, echr)), "European Convention on Human Rights", NA),
ecthr_in_paragraph_dict = ifelse(any(str_detect(text, "\\bemd\\b")), "ECtHR Caselaw", NA),
constitution_in_paragraph_dict = ifelse(any(str_detect(text, grl)), "The Constitution of the Kingdom of Norway", NA),
crc_in_paragraph_dict = ifelse(any(str_detect(text, bkn)), "Convention on the Rights of the Child", NA),
iccpr_in_paragraph_dict = ifelse(
any(str_detect(text, spn)),
"International Covenant on Civil and Political Rights",
NA
)
) %>%
ungroup()
Get references from codebook for the coding of overlapping laws
text_links_summary <- text_links_summary %>%
mutate(dictionary_references_echr = str_extract_all(text, paste(echr, collapse = "|")),
dictionary_references_crc = str_extract_all(text, paste(bkn, collapse = "|")),
dictionary_references_iccpr = str_extract_all(text, paste(spn, collapse = "|")),
dictionary_references_grl = str_extract_all(text, paste(grl, collapse = "|")))
dictionary_references_echr <- text_links_summary %>%
unnest(dictionary_references_echr) %>%
select(published, dictionary_references_echr) %>%
group_by(published) %>%
summarise(laws_echr = str_flatten(dictionary_references_echr, collapse = ", ", na.rm = T))
dictionary_references_crc <- text_links_summary %>%
unnest(dictionary_references_crc) %>%
select(published, dictionary_references_crc) %>%
group_by(published) %>%
summarise(laws_crc = str_flatten(dictionary_references_crc, collapse = ", ", na.rm = T))
dictionary_references_iccpr <- text_links_summary %>%
unnest(dictionary_references_iccpr) %>%
select(published, dictionary_references_iccpr) %>%
group_by(published) %>%
summarise(laws_iccpr = str_flatten(dictionary_references_iccpr, collapse = ", ", na.rm = T))
dictionary_references_grl <- text_links_summary %>%
unnest(dictionary_references_grl) %>%
select(published, dictionary_references_grl) %>%
group_by(published) %>%
summarise(laws_grl = str_flatten(dictionary_references_grl, collapse = ", ", na.rm = T))
# text_links_summary %>%
# select(published, article, ditionary_references_echr) %>%
# unnest(ditionary_references_echr, keep_empty = T) %>%
# filter(published == "HR-2016-389-A")
# paragraphs_text_ref <- paragraphs_text %>%
# distinct(case.x, par_paragraph, .keep_all = T) %>%
# mutate(
# echr_dict = str_extract_all(text, echr),
# ecthr_dict = str_extract_all(text, "\\bemd\\b"),
# constitution_dict = str_extract_all(text, grl),
# crc_dict = str_extract_all(text, bkn),
# iccpr_dict = str_extract_all(text, spn))
3.4 Law attributes
3.4.1 Number of unique FR laws
We create a variable that counts the number of unique FR laws cited
in the decision that we use as a control in the analysis.
# All laws cited in decision
laws <- text_links_summary %>%
group_by(case.x, published, section_detailed2) %>%
summarise(laws = str_flatten(article, collapse = ", ", na.rm = T))
# Unique fr laws cited in decision
article_uniqe <- text_links_summary %>%
filter(law_name == "European Convention on Human Rights"| law_name == "The Constitution of the Kingdom of Norway"| law_name == "Convention on the Rights of the Child"| law_name == "International Covenant on Civil and Political Rights") %>%
group_by(case.x, article) %>%
distinct(case.x, article) %>%
group_by(case.x) %>%
count() %>%
rename(unique_frlaw_articles = n)
3.4.2 Overlapping FR laws per table 2 in manuscript
We create variables that that identify cases with overlapping FR laws
and that we use in the analysis shown in figure 4 in the main
manuscript
laws_overlap <- text_links_summary %>%
filter(law_name == "European Convention on Human Rights"| law_name == "The Constitution of the Kingdom of Norway"| law_name == "Convention on the Rights of the Child"| law_name == "International Covenant on Civil and Political Rights") %>%
group_by(case.x, published) %>%
summarise(laws = str_flatten(article, collapse = ", ", na.rm = T))
laws_overlap <- laws_overlap %>%
left_join(dictionary_references_echr, by = "published") %>%
left_join(dictionary_references_crc, by = "published") %>%
left_join(dictionary_references_iccpr, by = "published") %>%
left_join(dictionary_references_grl, by = "published")
Overlap right to privacy
laws_overlap <- laws_overlap %>%
mutate(
overlap_privacy = case_when(
str_detect(laws, "emkn a8") &
str_detect(laws, "rpost2014_grl102") |
str_detect(laws, "emkn a8") &
str_detect(laws, "spn a17") |
str_detect(laws, "rpost2014_grl102") &
str_detect(laws, "spn a17") |
str_detect(laws_echr, "8") & str_detect(laws_iccpr, "17")|
str_detect(laws_echr, "8") & str_detect(laws_grl, "102")|
str_detect(laws_grl, "102") & str_detect(laws_iccpr, "17") ~ "Right to privacy"
)
)
overlap freedom of expression
laws_overlap <- laws_overlap %>%
mutate(
overlap_expression = case_when(
str_detect(laws, "emkn a10") &
str_detect(laws, "r_grl100") |
str_detect(laws, "emkn a10") &
str_detect(laws, "spn a19") |
str_detect(laws, "r_grl100") &
str_detect(laws, "spn a19") |
str_detect(laws_echr, "10") & str_detect(laws_iccpr, "19")|
str_detect(laws_echr, "10") & str_detect(laws_grl, "100")|
str_detect(laws_grl, "100") & str_detect(laws_iccpr, "19")
~ "Freedom of Expression"
))
overlap freeom of assembly
laws_overlap <- laws_overlap %>%
mutate(
overlap_assembly = case_when(
str_detect(laws, "emkn a11") &
str_detect(laws, "rpost2014_grl101") |
str_detect(laws, "emkn a11") &
str_detect(laws, "spn a22") |
str_detect(laws, "rpost2014_grl101") &
str_detect(laws, "spn a22")|
str_detect(laws_echr, "11") & str_detect(laws_iccpr, "22")|
str_detect(laws_echr, "11") & str_detect(laws_grl, "101")|
str_detect(laws_grl, "101") & str_detect(laws_iccpr, "22") ~ "Freedom of Assembly"))
overlap childrens rights
laws_overlap <- laws_overlap %>%
mutate(
overlap_children = case_when(
# str_detect(laws, "rpost2014_grl102") &
# str_detect(laws, "bkn") |
str_detect(laws, "rpost2014_grl104") &
str_detect(laws, "bkn")|
# !is.na(laws_crc) & str_detect(laws_grl, "102")|
!is.na(laws_crc) & str_detect(laws_grl, "104") ~ "Children"
)
)
overlap physical integrity
laws_overlap <- laws_overlap %>%
mutate(
overlap_life = case_when(
str_detect(laws, "emkn a2") & str_detect(laws, "rpost2014_grl93")|
str_detect(laws, "emkn a2") & str_detect(laws, "spn a6")|
str_detect(laws, "rpost2014_grl93") & str_detect(laws, "spn a6")|
str_detect(laws, "emkn a3") & str_detect(laws, "spn a7") |
str_detect(laws_echr, "emk artikkel 2") & str_detect(laws_iccpr, "artikkel 6")|
str_detect(laws_echr, "emk artikkel 2") & str_detect(laws_grl, "93")|
str_detect(laws_grl, "93") & str_detect(laws_iccpr, "artikkel 6")|
str_detect(laws_echr, "emk artikkel 3") & str_detect(laws_iccpr, "artikkel 7") ~ "Physical integrity"))
overlap liberty
laws_overlap <- laws_overlap %>%
mutate(
overlap_liberty = case_when(
str_detect(laws, "emkn a5") &
str_detect(laws, "rpost2014_grl94")|
str_detect(laws, "emkn a5") &
str_detect(laws, "r_grl99")|
str_detect(laws, "emkn a5") &
str_detect(laws, "spn a9")|
str_detect(laws, "rpost2014_grl94") &
str_detect(laws, "spn a9")|
str_detect(laws, "r_grl99") &
str_detect(laws, "spn a9")|
str_detect(laws_echr, "5") & str_detect(laws_iccpr, "9")|
str_detect(laws_echr, "5") & str_detect(laws_grl, "94")|
str_detect(laws_grl, "94") & str_detect(laws_iccpr, "5") ~ "Liberty"))
overlap fair trail
laws_overlap <- laws_overlap %>%
mutate(
overlap_fairtrial = case_when(
str_detect(laws, "emkn a6") &
str_detect(laws, "rpost2014_grl95")|
str_detect(laws, "emkn a6") &
str_detect(laws, "spn a14")|
str_detect(laws, "rpost2014_grl95") &
str_detect(laws, "spn a14")|
str_detect(laws_echr, "6") & str_detect(laws_iccpr, "14")|
str_detect(laws_echr, "6") & str_detect(laws_grl, "95")|
str_detect(laws_grl, "95") & str_detect(laws_iccpr, "14") ~ "Fair trial"))
overlap no punishment witout law
laws_overlap <- laws_overlap %>%
mutate(
overlap_nopunishmentwithoutlaw = case_when(
str_detect(laws, "emkn a7") &
str_detect(laws, "rpost2014_grl96")|
str_detect(laws, "emkn a7") &
str_detect(laws, "rpre2014_grl96")|
str_detect(laws, "emkn a7") &
str_detect(laws, "spn a9")|
str_detect(laws, "rpre2014_grl96") &
str_detect(laws, "spn a9")|
str_detect(laws, "rpost2014_grl96") &
str_detect(laws, "spn a9") ~ "No punishment without law"),
overlap_nopunishmentwithoutlaw2 = case_when(
str_detect(laws_echr, "emk artikkel 7") & str_detect(laws_iccpr, "9")|
str_detect(laws_echr, "emk artikkel 7") & str_detect(laws_grl, "96")|
str_detect(laws_grl, "96") & str_detect(laws_iccpr, "9") ~ "No punishment without law"))
overlap ex post facto and private property
laws_overlap <- laws_overlap %>%
mutate(
overlap_exostfacto_property = case_when(
str_detect(laws, "emkn p1 a1") & str_detect(laws, "r_grl97")|
str_detect(laws, "emkn p1 a1") & str_detect(laws, "r_grl105")|
str_detect(laws_echr, "p1|protokoll 1") & str_detect(laws_grl, "97")|
str_detect(laws_echr, "p1|protokoll 1") & str_detect(laws_grl, "105") ~ "Ex post facto & property"))
overlap guarantee of equality
laws_overlap <- laws_overlap %>%
mutate(
overlap_equality = case_when(
str_detect(laws, "emkn a14") &
str_detect(laws, "rpost2014_grl98")|
str_detect(laws, "emkn a14") &
str_detect(laws, "spn a26")|
str_detect(laws, "rpost2014_grl98") &
str_detect(laws, "spn a26") |
str_detect(laws_echr, "14") & str_detect(laws_iccpr, "26")|
str_detect(laws_echr, "14") & str_detect(laws_grl, "98")|
str_detect(laws_grl, "98") & str_detect(laws_iccpr, "26") ~ "Guarantee of equality"))
overlap culture and language
laws_overlap <- laws_overlap %>%
mutate(
overlap_culture = case_when(
str_detect(laws, "spn a27") &
str_detect(laws, "rpost2014_grl108") | str_detect(laws_iccpr, "27") & str_detect(laws_grl, "108")
~ "Culture and language"))
overlap freedom of movement
laws_overlap <- laws_overlap %>%
mutate(
overlap_freedom_of_movement = case_when(
str_detect(laws, "emkn p4 a3") &
str_detect(laws, "rpost2014_grl106")|
str_detect(laws, "emkn p4 a3") &
str_detect(laws, "spn a12")|
str_detect(laws, "rpost2014_grl106") &
str_detect(laws, "spn a12")|
str_detect(laws_echr, "protokoll 4 artikkel 3") & str_detect(laws_iccpr, "12")|
str_detect(laws_echr, "protokoll 4 artikkel 3") & str_detect(laws_grl, "106")|
str_detect(laws_grl, "106") & str_detect(laws_iccpr, "12")
~ "Freedom of movement"))
3.4.3 Outcome variables used in analysis
For each relevant FR law, we count the number of paragraphs that cite
or mention that specific FR law. The difference between ECHR and
ECHR_small is that the former includes references and mentions of ECtHR,
while the latter does not.
data <- paragraphs_text %>%
group_by(case.x, section_detailed2) %>%
mutate(
ECHR = sum(
echr_in_paragraph == "European Convention on Human Rights" |
ecthr_in_paragraph == "ECtHR Caselaw" |
echr_in_paragraph_dict == "European Convention on Human Rights" |
ecthr_in_paragraph_dict == "ECtHR Caselaw",
na.rm = T
),
ECHR_small = sum(
echr_in_paragraph == "European Convention on Human Rights" |
echr_in_paragraph_dict == "European Convention on Human Rights",
na.rm = T
),
ECtHR = sum(
ecthr_in_paragraph == "ECtHR Caselaw" |
ecthr_in_paragraph_dict == "ECtHR Caselaw",
na.rm = T
),
CON = sum(
constitution_in_paragraph == "The Constitution of the Kingdom of Norway" |
constitution_in_paragraph_dict == "The Constitution of the Kingdom of Norway",
na.rm = T
),
CRC = sum(
crc_in_paragraph == "Convention on the Rights of the Child" |
crc_in_paragraph_dict == "Convention on the Rights of the Child"|
un_gc_crc_in_paragraph == "UN General Comment The Committee on the Rights of the Child",
na.rm = T
),
CRC_small = sum(
crc_in_paragraph == "Convention on the Rights of the Child",
na.rm = T
),
CRC_caselaw = sum(
un_gc_crc_in_paragraph == "UN General Comment The Committee on the Rights of the Child",
na.rm = T
),
ICCPR = sum(
iccpr_in_paragraph == "International Covenant on Civil and Political Rights" |
iccpr_caselaw_in_paragraph == "ICCPR Caselaw"|
un_gc_ccpr_in_paragraph == "UN General Comment The Committee on Civil and Political Rights"|
iccpr_in_paragraph_dict == "International Covenant on Civil and Political Rights",
na.rm = T
),
ICCPR_small = sum(
iccpr_in_paragraph == "International Covenant on Civil and Political Rights" |
iccpr_in_paragraph_dict == "International Covenant on Civil and Political Rights",
na.rm = T
),
ICCPR_caselaw = sum(
iccpr_caselaw_in_paragraph == "ICCPR Caselaw"|
un_gc_ccpr_in_paragraph == "UN General Comment The Committee on Civil and Political Rights",
na.rm = T
),
# fr_law_all = sum(
# echr_in_paragraph == "European Convention on Human Rights" |
# ecthr_in_paragraph == "ECtHR Caselaw" |
# echr_in_paragraph_dict == "European Convention on Human Rights" |
# ecthr_in_paragraph_dict == "ECtHR Caselaw" |
# constitution_in_paragraph == "The Constitution of the Kingdom of Norway" |
# constitution_in_paragraph_dict == "The Constitution of the Kingdom of Norway" |
# crc_in_paragraph == "Convention on the Rights of the Child" |
# crc_in_paragraph_dict == "Convention on the Rights of the Child" |
# un_gc_crc_in_paragraph == "UN General Comment The Committee on the Rights of the Child",
# iccpr_in_paragraph == "International Covenant on Civil and Political Rights" |
# iccpr_in_paragraph_dict == "International Covenant on Civil and Political Rights"|
# iccpr_caselaw_in_paragraph == "ICCPR Caselaw"|
# un_gc_ccpr_in_paragraph == "UN General Comment The Committee on Civil and Political Rights"|
# un_gc_cescr_in_paragraph == "UN General Comment The Committee on Economic, Social and Cultural Rights"|
# un_gc_crpd_in_paragraph == "UN General Comment The Committee on the Rights of Persons with Disabilities"|
# cedaw_in_paragraph == "Convention on the Elimination of All Forms of Discrimination against Women"|
# icescr_in_paragraph == "International Covenant on Economic, Social and Cultural Rights",
# na.rm = T
# ),
n_paragraphs = n()
) %>%
distinct(case.x, section_detailed2, .keep_all = T)
3.4.4 FR law mentioned in main question
data <- data %>%
group_by(case.x) %>%
mutate(
# frlaw_mainQ = case_when(section_detailed2 == "Main question" & fr_law_all > 0 ~ "FR law in main Q",
# section_detailed2 == "Main question" & fr_law_all == 0 ~ "FR law not in main Q"),
echr_mainQ = case_when(section_detailed2 == "Main question" & ECHR > 0 ~ "ECHR in main Q",
section_detailed2 == "Main question" & ECHR == 0 ~ "ECHR not in main Q"),
constitution_mainQ = case_when(section_detailed2 == "Main question" & CON > 0 ~ "CON in main Q",
section_detailed2 == "Main question" & CON == 0 ~ "CON not in main Q"),
crc_mainQ = case_when(section_detailed2 == "Main question" & CRC > 0 ~ "CRC in main Q",
section_detailed2 == "Main question" & CRC == 0 ~ "CRC not in main Q"),
iccpr_mainQ = case_when(section_detailed2 == "Main question" & ICCPR > 0 ~ "ICCPR in main Q",
section_detailed2 == "Main question" & ICCPR == 0 ~ "ICCPR not in main Q")
) %>%
fill(
#frlaw_mainQ,
echr_mainQ, constitution_mainQ, crc_mainQ, iccpr_mainQ)
Keep the variables we are using
data <- data %>%
select(
case.x,
Dato,
par_judge,
par_text,
section_detailed2,
ECHR,
ECHR_small,
ECtHR,
CON,
CRC,
CRC_small,
CRC_caselaw,
ICCPR,
ICCPR_small,
ICCPR_caselaw,
echr_mainQ, constitution_mainQ, crc_mainQ, iccpr_mainQ,
n_paragraphs,
type_of_panel,
dissent_case,
type_of_law,
type_of_decision,
hr_number,
rt_number,
published,
plenary_impartiality
)
Merge data with the law attribute-variables
data <- data %>%
inner_join(laws, by = c("case.x", "section_detailed2"))
data <- data %>%
inner_join(article_uniqe, by = c("case.x"))
data <- data %>%
inner_join(laws_overlap, by = c("case.x"))
Pivot data from wide to long
data_long <- data %>%
pivot_longer(
cols = c("ECHR", "ECHR_small", "ECtHR",
"CON",
"CRC", "CRC_small", "CRC_caselaw",
"ICCPR", "ICCPR_small", "ICCPR_caselaw",
#"fr_law_all"
),
names_to = "FR_law",
values_to = "N"
)
Remove 0’s, keep only opinions (i.e., main opinions, dissenting
opinions)
data_long_opinions <- data_long %>%
filter(N > 0) %>%
group_by(case.x) %>%
mutate(fr_law_name = case_when(FR_law == "ECHR" ~ "ECHR",
FR_law == "CON" ~ "CON",
FR_law == "CRC" ~ "CRC",
FR_law == "ICCPR" ~ "ICCPR"),
n_fr_laws = n_distinct(fr_law_name, na.rm = T)) %>%
filter(section_detailed2 == "Main opinion"| str_detect(section_detailed2, "Dissenting")) %>%
ungroup()
#Figure 1 in manuscript. Increasing FR law complexity in SCON
#Main dataset
case_data <- data_long_opinions %>%
filter(is.na(plenary_impartiality)) %>%
group_by(case.x, FR_law) %>%
mutate(N = sum(N),
n_paragraphs = sum(n_paragraphs)) %>%
distinct(case.x, .keep_all = T) %>%
mutate(prop_para = (N/n_paragraphs)*100)
ANALYSIS
h1_case_data <- case_data %>%
mutate(complexity = ifelse(n_fr_laws >1, "Multiple FR laws", "One FR law"),
complexity = factor(complexity,
levels = c("One FR law", "Multiple FR laws")))
h1_case_data <- h1_case_data %>%
mutate(FR_law = factor(FR_law,
levels = c("CON", "CRC", "ICCPR", "ECHR")) )
# H1
lm1 <- lm(prop_para ~ FR_law, data = h1_case_data)
lm2 <- lm(prop_para ~ FR_law + complexity + type_of_law + type_of_panel + type_of_decision + dissent_case, data = h1_case_data)
#lm2 <- lm(prop_para ~ FR_law*complexity, data = h1_case_data)
fr_predictions <-
predictions(lm1, type = "response") %>%
group_by(FR_law) %>%
summarize(across(c(estimate, conf.low, conf.high), mean))
fr_contrast1 <- avg_comparisons(lm1, variables = list(FR_law = "pairwise")) %>%
mutate(test = case_when(
contrast == "mean(CRC) - mean(CON)" ~ "H3 UN < CON",
contrast == "mean(ECHR) - mean(CON)" ~ "H2 ECHR > CON",
contrast == "mean(ECHR) - mean(CRC)" ~ "H1 ECHR > UN",
contrast == "mean(ICCPR) - mean(CON)" ~ "H3 UN < CON",
contrast == "mean(ECHR) - mean(ICCPR)" ~ "H1 ECHR > UN"
))
fr_contrast1 <- fr_contrast1 %>%
mutate(nicelab = case_when(
contrast == "mean(CRC) - mean(CON)" ~ "CRC - CON",
contrast == "mean(ECHR) - mean(CON)" ~ "ECHR - CON",
contrast == "mean(ECHR) - mean(CRC)" ~ "ECHR - CRC",
contrast == "mean(ICCPR) - mean(CON)" ~ "ICCPR - CON",
contrast == "mean(ECHR) - mean(ICCPR)" ~ "ECHR - ICCPR"
))
fr_contrast1 <- fr_contrast1 %>%
filter(!is.na(test))
fr_predictions2 <-
predictions(lm2, type = "response") %>%
group_by(FR_law, complexity) %>%
summarize(across(c(estimate, conf.low, conf.high), mean))
fr_contrast2 <- avg_comparisons(lm2, variable = list(FR_law = "pairwise"), by = "complexity") %>%
mutate(test = case_when(
contrast == "mean(CRC) - mean(CON)" ~ "H2c UN < CON",
contrast == "mean(ECHR) - mean(CON)" ~ "H2b ECHR > CON",
contrast == "mean(ECHR) - mean(CRC)" ~ "H2a ECHR > UN",
contrast == "mean(ICCPR) - mean(CON)" ~ "H2c UN < CON",
contrast == "mean(ICCPR) - mean(ECHR)" ~ "H2a UN < ECHR"
),
contrast = case_when(
contrast == "mean(CRC) - mean(CON)" ~ "CRC - CON",
contrast == "mean(ECHR) - mean(CON)" ~ "ECHR - CON",
contrast == "mean(ECHR) - mean(CRC)" ~ "ECHR - CRC",
contrast == "mean(ICCPR) - mean(CON)" ~ "ICCPR - CON",
contrast == "mean(ICCPR) - mean(ECHR)" ~ "ICCPR - ECHR"
)) %>%
filter(!is.na(test))
modelsummary::modelsummary(
list(lm1, lm2), vcov = ~ case.x,
stars = TRUE,
output = "lm.docx")
# frlevels <-
# ggplot() + geom_point(
# data = subset(
# h2_case_data,
# FR_law == "ECHR" |
# FR_law == "CON" | FR_law == "CRC" | FR_law == "ICCPR"
# ),
# aes(x = FR_law, y = prop_para, colour = FR_law),
# alpha = 0.7
# ) +
# scale_colour_manual(values = c("#CC6666", "#9999CC", "#66CC99", "#E69F00"))
#
# frlevels <- frlevels +
# geom_pointrange(data = fr_predictions, aes(x = FR_law, y = estimate, ymin = conf.low, ymax = conf.high), shape = 21, colour = "black", fill = "white", inherit.aes = FALSE) +
# #scale_y_continuous(limits = c(0, 1)) +
# #facet_wrap(~ n_fr_laws) +
# #coord_flip() +
# labs(y = "Predicted proportion", x = "Type of FR law") +
# theme_minimal() +
# theme(legend.position = "none")
h1_desc <-
h1_case_data %>%
filter(!is.na(FR_law)) %>%
ggplot(aes(x = FR_law, y = prop_para)) +
geom_half_point(aes(color = FR_law),
transformation = position_quasirandom(width = 0.1),
side = "l", size = 2, alpha = 0.6) +
geom_half_boxplot(aes(fill = FR_law), side = "r", alpha = 0.6) +
scale_y_continuous(limits = c(0, 100)) +
scale_fill_viridis_d(option = "plasma", end = 0.8) +
scale_color_viridis_d(option = "plasma", end = 0.8) +
scale_x_discrete(labels = label_wrap(10)) +
facet_wrap (~ complexity) +
#coord_flip() +
guides(color = "none", fill = "none") +
labs(x = "", y = "Paragraphs(%)\nengaging with FR laws") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
#plot.title = element_text(face = "bold"),
#axis.title = element_text(face = "bold"),
strip.text = element_text(face = "bold"),
axis.text = element_text(colour = "black",
size = 12),
strip.text.x = element_text(size = 12),
strip.text.y = element_text(size = 12),
axis.text.x = element_text(size = 12),
axis.text.y = element_text(size = 12),
strip.background = element_rect(fill = "grey80", color = NA))
frcontrasts <- ggplot(fr_contrast1, aes(x = nicelab, y = estimate, ymin = conf.low, ymax = conf.high)) +
geom_pointrange() +
geom_hline(yintercept = 0, colour = "grey") +
#scale_y_continuous(limits = c(0, 0.5)) +
facet_wrap(~ test, scales = "free_y", ncol = 1) +
coord_flip() +
labs(y = "Difference in paragraphs(%)\nengaging with FR laws", x = "") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
#plot.title = element_text(face = "bold"),
#axis.title = element_text(face = "bold"),
strip.text = element_text(face = "bold"),
axis.text = element_text(colour = "black",
size = 12),
strip.text.x = element_text(size = 12),
strip.text.y = element_text(size = 12),
axis.text.x = element_text(size = 12),
axis.text.y = element_text(size = 12),
strip.background = element_rect(fill = "grey80", color = NA))
median(h1_case_data$prop_para[which(h1_case_data$FR_law == "ECHR")])
## [1] 18.51852
# ECHR
median(h1_case_data$prop_para[which(h1_case_data$FR_law == "ECHR" & h1_case_data$complexity == "One FR law")])
## [1] 18.51852
median(h1_case_data$prop_para[which(h1_case_data$FR_law == "ECHR" & h1_case_data$complexity == "Multiple FR laws")])
## [1] 18.39139
# CON
median(h1_case_data$prop_para[which(h1_case_data$FR_law == "CON" & h1_case_data$complexity == "One FR law")])
## [1] 13.04348
median(h1_case_data$prop_para[which(h1_case_data$FR_law == "CON" & h1_case_data$complexity == "Multiple FR laws")])
## [1] 5.882353
# CON
median(h1_case_data$prop_para[which(h1_case_data$FR_law == "CRC" & h1_case_data$complexity == "One FR law")])
## [1] 18.42105
median(h1_case_data$prop_para[which(h1_case_data$FR_law == "CRC" & h1_case_data$complexity == "Multiple FR laws")])
## [1] 4.819977
# CON
median(h1_case_data$prop_para[which(h1_case_data$FR_law == "ICCPR" & h1_case_data$complexity == "One FR law")])
## [1] 14.28571
median(h1_case_data$prop_para[which(h1_case_data$FR_law == "ICCPR" & h1_case_data$complexity == "Multiple FR laws")])
## [1] 4.188596
median(h1_case_data$prop_para[which(h1_case_data$FR_law == "ECHR" & h1_case_data$complexity == "One FR law")])
## [1] 18.51852
H1 Constituional amendment
H1 Enlarged panel
type_panel <- h1_case_data %>%
filter(!is.na(FR_law)) %>%
ggplot(aes(x = FR_law, y = prop_para)) +
geom_half_point(aes(color = FR_law),
transformation = position_quasirandom(width = 0.1),
side = "l", size = 1, alpha = 0.6) +
geom_half_boxplot(aes(fill = FR_law), side = "r", alpha = 0.7) +
scale_y_continuous(limits = c(0, 100)) +
scale_fill_viridis_d(option = "plasma", end = 0.8) +
scale_color_viridis_d(option = "plasma", end = 0.8) +
scale_x_discrete(labels = label_wrap(10)) +
facet_wrap (~ type_of_panel) +
#coord_flip() +
guides(color = "none", fill = "none") +
labs(x = "", y = "Paragraphs(%)\nengaging with FR laws") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
#plot.title = element_text(face = "bold"),
#axis.title = element_text(face = "bold"),
strip.text = element_text(face = "bold"),
strip.background = element_rect(fill = "grey80", color = NA))
H1 types of decision
type_decision <- h1_case_data %>%
filter(!is.na(FR_law)) %>%
ggplot(aes(x = FR_law, y = prop_para)) +
geom_half_point(aes(color = FR_law),
transformation = position_quasirandom(width = 0.1),
side = "l", size = 1, alpha = 0.7) +
geom_half_boxplot(aes(fill = FR_law), side = "r", alpha = 0.6) +
scale_y_continuous(limits = c(0, 100)) +
scale_fill_viridis_d(option = "plasma", end = 0.8) +
scale_color_viridis_d(option = "plasma", end = 0.8) +
scale_x_discrete(labels = label_wrap(10)) +
facet_wrap (~ type_of_decision) +
#coord_flip() +
guides(color = "none", fill = "none") +
labs(x = "", y = "Paragraphs(%)\nengaging with FR laws") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
#plot.title = element_text(face = "bold"),
#axis.title = element_text(face = "bold"),
strip.text = element_text(face = "bold"),
strip.background = element_rect(fill = "grey80", color = NA))
H1 civil and criminal
type_law <- h1_case_data %>%
filter(!is.na(FR_law)) %>%
ggplot(aes(x = FR_law, y = prop_para)) +
geom_half_point(aes(color = FR_law),
transformation = position_quasirandom(width = 0.1),
side = "l", size = 1, alpha = 0.7) +
geom_half_boxplot(aes(fill = FR_law), side = "r", alpha = 0.6) +
scale_y_continuous(limits = c(0, 100)) +
scale_fill_viridis_d(option = "plasma", end = 0.8) +
scale_color_viridis_d(option = "plasma", end = 0.8) +
scale_x_discrete(labels = label_wrap(10)) +
facet_wrap (~ type_of_law) +
#coord_flip() +
guides(color = "none", fill = "none") +
labs(x = "", y = "Paragraphs(%)\nengaging with FR laws") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
#plot.title = element_text(face = "bold"),
#axis.title = element_text(face = "bold"),
strip.text = element_text(face = "bold"),
strip.background = element_rect(fill = "grey80", color = NA))
DISSENT
h1_case_data %>%
filter(!is.na(FR_law)) %>%
ggplot(aes(x = FR_law, y = prop_para)) +
geom_half_point(aes(color = FR_law),
transformation = position_quasirandom(width = 0.1),
side = "l", size = 1, alpha = 0.7) +
geom_half_boxplot(aes(fill = FR_law), side = "r", alpha = 0.6) +
scale_y_continuous(limits = c(0, 100)) +
scale_fill_viridis_d(option = "plasma", end = 0.8) +
scale_color_viridis_d(option = "plasma", end = 0.8) +
scale_x_discrete(labels = label_wrap(10)) +
facet_wrap (~ dissent_case) +
#coord_flip() +
guides(color = "none", fill = "none") +
labs(x = "", y = "Paragraphs(%) engaging with FR laws") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
#plot.title = element_text(face = "bold"),
#axis.title = element_text(face = "bold"),
strip.text = element_text(face = "bold"),
strip.background = element_rect(fill = "grey80", color = NA))
H1 Overlapping FR laws
1. Right to life
overlap1 <- h1_case_data %>%
filter(overlap_life == "Physical integrity",
!is.na(FR_law)) %>%
mutate(panel_label = "Right to Life") %>%
ggplot(aes(x = FR_law, y = prop_para)) +
geom_half_point(aes(color = FR_law),
transformation = position_quasirandom(width = 0.1),
side = "l", size = 2, alpha = 0.6) +
geom_half_boxplot(aes(fill = FR_law), side = "r", alpha = 0.7) +
scale_y_continuous(limits = c(0, 100)) +
scale_fill_viridis_d(option = "plasma", end = 0.8) +
scale_color_viridis_d(option = "plasma", end = 0.8) +
scale_x_discrete(labels = label_wrap(10)) +
#ggrepel::geom_text_repel(min.segment.length = 0, size = 2) +
facet_wrap (~ panel_label) +
#coord_flip() +
guides(color = "none", fill = "none") +
labs(x = "", y = "Paragraphs(%)\nengaging with FR laws") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
#plot.title = element_text(face = "bold"),
#axis.title = element_text(face = "bold"),
strip.text = element_text(face = "bold"),
strip.background = element_rect(fill = "grey80", color = NA))
2. Right to liberty
overlap2 <- h1_case_data %>%
filter(overlap_liberty == "Liberty",
!is.na(FR_law)) %>%
mutate(panel_label = "Right to Liberty") %>%
ggplot(aes(x = FR_law, y = prop_para)) +
geom_half_point(aes(color = FR_law),
transformation = position_quasirandom(width = 0.1),
side = "l", size = 2, alpha = 0.6) +
geom_half_boxplot(aes(fill = FR_law), side = "r", alpha = 0.7) +
scale_y_continuous(limits = c(0, 100)) +
scale_fill_viridis_d(option = "plasma", end = 0.8) +
scale_color_viridis_d(option = "plasma", end = 0.8) +
scale_x_discrete(labels = label_wrap(10)) +
facet_wrap (~ panel_label) +
#coord_flip() +
guides(color = "none", fill = "none") +
labs(x = "", y = "") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
axis.text.y=element_blank(),
axis.ticks.y=element_blank(),
strip.text = element_text(face = "bold"),
strip.background = element_rect(fill = "grey80", color = NA))
3. Fair trial
overlap3 <- h1_case_data %>%
filter(overlap_fairtrial == "Fair trial",
!is.na(FR_law),
FR_law != "CRC") %>%
mutate(panel_label = "The Right to a Fair Trial") %>%
ggplot(aes(x = FR_law, y = prop_para)) +
geom_half_point(aes(color = FR_law),
transformation = position_quasirandom(width = 0.1),
side = "l", size = 2, alpha = 0.6) +
geom_half_boxplot(aes(fill = FR_law), side = "r", alpha = 0.7) +
scale_y_continuous(limits = c(0, 100)) +
scale_fill_viridis_d(option = "plasma", end = 0.8) +
scale_color_viridis_d(option = "plasma", end = 0.8) +
scale_x_discrete(labels = label_wrap(10)) +
facet_wrap (~ panel_label) +
#coord_flip() +
guides(color = "none", fill = "none") +
labs(x = "", y = "Paragraphs(%)\nengaging with FR laws") +
theme_minimal() +
labs(x = "", y = "") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
axis.text.y=element_blank(),
axis.ticks.y=element_blank(),
strip.text = element_text(face = "bold"),
strip.background = element_rect(fill = "grey80", color = NA))
4. No punishment without law
overlap4 <- h1_case_data %>%
filter(overlap_nopunishmentwithoutlaw == "No punishment without law",
!is.na(FR_law)) %>%
mutate(panel_label = "No Punishment\nWithout Law") %>%
ggplot(aes(x = FR_law, y = prop_para)) +
geom_half_point(aes(color = FR_law),
transformation = position_quasirandom(width = 0.1),
side = "l", size = 2, alpha = 0.6) +
geom_half_boxplot(aes(fill = FR_law), side = "r", alpha = 0.7) +
scale_y_continuous(limits = c(0, 100)) +
scale_fill_viridis_d(option = "plasma", end = 0.8) +
scale_color_viridis_d(option = "plasma", end = 0.8) +
scale_x_discrete(labels = label_wrap(10)) +
facet_wrap (~ panel_label) +
#coord_flip() +
guides(color = "none", fill = "none") +
labs(x = "", y = "Paragraphs(%)\nengaging with FR laws") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
#plot.title = element_text(face = "bold"),
#axis.title = element_text(face = "bold"),
strip.text = element_text(face = "bold"),
strip.background = element_rect(fill = "grey80", color = NA))
5. General Guarantee of Equality
overlap5 <- h1_case_data %>%
filter(overlap_equality == "Guarantee of equality",
!is.na(FR_law)) %>%
mutate(panel_label = "General Guarantee\nof Equality") %>%
ggplot(aes(x = FR_law, y = prop_para)) +
geom_half_point(aes(color = FR_law),
transformation = position_quasirandom(width = 0.1),
side = "l", size = 2, alpha = 0.6) +
geom_half_boxplot(aes(fill = FR_law), side = "r", alpha = 0.7) +
scale_y_continuous(limits = c(0, 100)) +
scale_fill_viridis_d(option = "plasma", end = 0.8) +
scale_color_viridis_d(option = "plasma", end = 0.8) +
scale_x_discrete(labels = label_wrap(10)) +
facet_wrap (~ panel_label) +
#coord_flip() +
guides(color = "none", fill = "none") +
labs(x = "", y = "Paragraphs(%)\nengaging with FR laws") +
theme_minimal() +
labs(x = "", y = "") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
axis.text.y=element_blank(),
axis.ticks.y=element_blank(),
strip.text = element_text(face = "bold"),
strip.background = element_rect(fill = "grey80", color = NA))
6. Freedom of expression
overlap6 <- h1_case_data %>%
filter(overlap_expression == "Freedom of Expression",
FR_law != "CRC") %>%
mutate(panel_label = "Freedom of\nExpression") %>%
ggplot(aes(x = FR_law, y = prop_para)) +
geom_half_point(aes(color = FR_law),
transformation = position_quasirandom(width = 0.1),
side = "l", size = 2, alpha = 0.6) +
geom_half_boxplot(aes(fill = FR_law), side = "r", alpha = 0.7) +
scale_y_continuous(limits = c(0, 100)) +
scale_fill_viridis_d(option = "plasma", end = 0.8) +
scale_color_viridis_d(option = "plasma", end = 0.8) +
scale_x_discrete(labels = label_wrap(10)) +
facet_wrap (~ panel_label) +
#coord_flip() +
guides(color = "none", fill = "none") +
labs(x = "", y = "Paragraphs(%)\nengaging with FR laws") +
theme_minimal() +
labs(x = "", y = "") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
axis.text.y=element_blank(),
axis.ticks.y=element_blank(),
strip.text = element_text(face = "bold"),
strip.background = element_rect(fill = "grey80", color = NA))
7. Freedom of Assembly
overlap7 <- h1_case_data %>%
filter(overlap_assembly == "Freedom of Assembly",
!is.na(FR_law)) %>%
mutate(panel_label = "Freedom of\nAssembly") %>%
ggplot(aes(x = FR_law, y = prop_para)) +
geom_half_point(aes(color = FR_law),
transformation = position_quasirandom(width = 0.1),
side = "l", size = 2, alpha = 0.6) +
geom_half_boxplot(aes(fill = FR_law), side = "r", alpha = 0.7) +
scale_y_continuous(limits = c(0, 100)) +
scale_fill_viridis_d(option = "plasma", end = 0.8) +
scale_color_viridis_d(option = "plasma", end = 0.8) +
scale_x_discrete(labels = label_wrap(10)) +
facet_wrap (~ panel_label) +
#coord_flip() +
guides(color = "none", fill = "none") +
labs(x = "", y = "Paragraphs(%)\nengaging with FR laws") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
#plot.title = element_text(face = "bold"),
#axis.title = element_text(face = "bold"),
strip.text = element_text(face = "bold"),
strip.background = element_rect(fill = "grey80", color = NA))
8. Right to privacy
overlap8 <- h1_case_data %>%
filter(overlap_privacy == "Right to privacy",
FR_law != "CRC") %>%
mutate(panel_label = "Right to\nPrivacy") %>%
ggplot(aes(x = FR_law, y = prop_para)) +
geom_half_point(aes(color = FR_law),
transformation = position_quasirandom(width = 0.1),
side = "l", size = 2, alpha = 0.6) +
geom_half_boxplot(aes(fill = FR_law), side = "r", alpha = 0.7) +
scale_y_continuous(limits = c(0, 100)) +
scale_fill_viridis_d(option = "plasma", end = 0.8) +
scale_color_viridis_d(option = "plasma", end = 0.8) +
scale_x_discrete(labels = label_wrap(10)) +
facet_wrap (~ panel_label) +
#coord_flip() +
guides(color = "none", fill = "none") +
labs(x = "", y = "Paragraphs(%)\nengaging with FR laws") +
theme_minimal() +
labs(x = "", y = "") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
axis.text.y=element_blank(),
axis.ticks.y=element_blank(),
strip.text = element_text(face = "bold"),
strip.background = element_rect(fill = "grey80", color = NA))
9. Children’s rights
overlap9 <- h1_case_data %>%
filter(overlap_children == "Children", !is.na(FR_law)) %>%
mutate(panel_label = "Children's\nRights") %>%
ggplot(aes(x = FR_law, y = prop_para)) +
geom_half_point(aes(color = FR_law),
transformation = position_quasirandom(width = 0.1),
side = "l", size = 2, alpha = 0.7) +
geom_half_boxplot(aes(fill = FR_law), side = "r", alpha = 0.7) +
scale_y_continuous(limits = c(0, 100)) +
scale_fill_viridis_d(option = "plasma", end = 0.8) +
scale_color_viridis_d(option = "plasma", end = 0.8) +
scale_x_discrete(labels = label_wrap(10)) +
facet_wrap (~ panel_label) +
#coord_flip() +
guides(color = "none", fill = "none") +
labs(x = "", y = "Paragraphs(%)\nengaging with FR laws") +
theme_minimal() +
labs(x = "", y = "") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
axis.text.y=element_blank(),
axis.ticks.y=element_blank(),
strip.text = element_text(face = "bold"),
strip.background = element_rect(fill = "grey80", color = NA))
10. Ex post facto and private property
overlap10 <- h1_case_data %>%
filter(overlap_exostfacto_property == "Ex post facto & property",
!is.na(FR_law)) %>%
mutate(panel_label = "Private Property &\nBan on\nEx Post Facto Laws") %>%
ggplot(aes(x = FR_law, y = prop_para)) +
geom_half_point(aes(color = FR_law),
transformation = position_quasirandom(width = 0.1),
side = "l", size = 2, alpha = 0.6) +
geom_half_boxplot(aes(fill = FR_law), side = "r", alpha = 0.7) +
scale_y_continuous(limits = c(0, 100)) +
scale_fill_viridis_d(option = "plasma", end = 0.8) +
scale_color_viridis_d(option = "plasma", end = 0.8) +
scale_x_discrete(labels = label_wrap(10)) +
facet_wrap (~ panel_label) +
#coord_flip() +
guides(color = "none", fill = "none") +
labs(x = "", y = "Paragraphs(%)\nengaging with FR laws") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
#plot.title = element_text(face = "bold"),
#axis.title = element_text(face = "bold"),
strip.text = element_text(face = "bold"),
strip.background = element_rect(fill = "grey80", color = NA))
11. Freedom of movement
overlap11 <- h1_case_data %>%
filter(overlap_freedom_of_movement == "Freedom of movement",
!is.na(FR_law),
FR_law != "CRC") %>%
mutate(panel_label = "Freedom\nof\nMovement") %>%
ggplot(aes(x = FR_law, y = prop_para)) +
geom_half_point(aes(color = FR_law),
transformation = position_quasirandom(width = 0.1),
side = "l", size = 2, alpha = 0.6) +
geom_half_boxplot(aes(fill = FR_law), side = "r", alpha = 0.7) +
scale_y_continuous(limits = c(0, 100)) +
scale_fill_viridis_d(option = "plasma", end = 0.8) +
scale_color_viridis_d(option = "plasma", end = 0.8) +
scale_x_discrete(labels = label_wrap(10)) +
facet_wrap (~ panel_label) +
#coord_flip() +
guides(color = "none", fill = "none") +
labs(x = "", y = "Paragraphs(%)\nengaging with FR laws") +
theme_minimal() +
labs(x = "", y = "") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
axis.text.y=element_blank(),
axis.ticks.y=element_blank(),
strip.text = element_text(face = "bold"),
strip.background = element_rect(fill = "grey80", color = NA))
12. Right to culture
overlap12 <- h1_case_data %>%
filter(overlap_culture == "Culture and language",
!is.na(FR_law),
FR_law != "CRC") %>%
mutate(panel_label = "Right\nto\nCulture") %>%
ggplot(aes(x = FR_law, y = prop_para)) +
geom_half_point(aes(color = FR_law),
transformation = position_quasirandom(width = 0.1),
side = "l", size = 2, alpha = 0.6) +
geom_half_boxplot(aes(fill = FR_law), side = "r", alpha = 0.7) +
scale_y_continuous(limits = c(0, 100)) +
scale_fill_viridis_d(option = "plasma", end = 0.8) +
scale_color_viridis_d(option = "plasma", end = 0.8) +
scale_x_discrete(labels = label_wrap(10)) +
facet_wrap (~ panel_label) +
#coord_flip() +
guides(color = "none", fill = "none") +
labs(x = "", y = "Paragraphs(%)\nengaging with FR laws") +
theme_minimal() +
labs(x = "", y = "") +
theme_minimal() +
theme(panel.grid.minor = element_blank(),
plot.background = element_rect(fill = "white", color = NA),
axis.text.y=element_blank(),
axis.ticks.y=element_blank(),
strip.text = element_text(face = "bold"),
strip.background = element_rect(fill = "grey80", color = NA))
