Rendered at 10:38 AM, Apr. 17, 2023
Ryan Thornburg (ryan.thornburg@unc.edu)
1 = Mention in passing (1 mention)
2 = Mention and includes some detail/discussion (2-3 mentions)
3 = In depth detail/discussion (more than 3 mentions or significant discussion of office/policy)
inc_unique_articles %>%
filter(type_of_prosecutor_mention %in% c("1","2","3")) %>%
group_by(type_of_prosecutor_mention) %>%
summarize (articles=n()) %>%
mutate(percentage = scales::percent(articles / sum(articles)))# A tibble: 3 × 3
type_of_prosecutor_mention articles percentage
<chr> <int> <chr>
1 1 739 45%
2 2 545 33%
3 3 356 22% ninc_unique_articles %>%
filter(type_of_candidate_mention %in% c("1","2","3")) %>%
group_by(type_of_candidate_mention) %>%
summarize (articles=n()) %>%
mutate(percentage = scales::percent(articles / sum(articles))) # A tibble: 3 × 3
type_of_candidate_mention articles percentage
<chr> <int> <chr>
1 1 146 45%
2 2 71 22%
3 3 111 34% 0 = neither is mentioned
1 = case-specific/particular prosecutor decision mentioned (describe in notes)
2 = general/office-wide prosecutor policy mentioned (describe in notes)
Note that we accepted multiple values for each article. So we have a total of 1,698 focuses for the 1,646 articles that mention an incumbent prosecutor.
inc_unique_articles %>%
mutate(code_1 = case_when(str_detect(primary_focus_of_article,"1") ~ 1,
TRUE ~ 0),
code_2 = case_when(str_detect(primary_focus_of_article,"2") ~ 1,
TRUE ~ 0),
code_3 = case_when(str_detect(primary_focus_of_article,"3") ~ 1,
TRUE ~ 0),
code_4 = case_when(str_detect(primary_focus_of_article,"4") ~ 1,
TRUE ~ 0),
code_5 = case_when(str_detect(primary_focus_of_article,"5") ~ 1,
TRUE ~ 0),
code_99 = case_when(str_detect(primary_focus_of_article,"99") ~ 1,
#NOTE: We had one article coded "0", which was not a valid code
str_detect(primary_focus_of_article,"0") ~ 1,
TRUE ~ 0),
) %>%
select(2,1, starts_with(c("primary","code_"))) %>%
select(1,2,starts_with("code_")) %>%
summarize(across(starts_with("code_"),sum)) %>%
rename(specific_crime_case = code_1,
category_crimes_cases = code_2,
prosecutor = code_3,
election = code_4,
justice_system_general = code_5,
other = code_99) %>%
pivot_longer(1:6, names_to = "primary_focus", values_to = "articles") %>%
arrange(desc(articles)) %>%
adorn_percentages("col") %>%
adorn_pct_formatting() %>%
datatable(options = list(dom='t'),
colnames = c('Focus', '% of Mentions'),
rownames = FALSE)A single article can discuss multiple types of crimes, so this shows the number of time each crime is mentioned. Use 1,646 articles as the denominator to determine the percentage of articles that discuss a certain crime.
inc_unique_articles %>%
mutate(
code_0 = case_when(str_detect(type_of_crime_discussed,"^0") ~ TRUE),
code_1 = case_when(str_detect(type_of_crime_discussed,"^1$|^1[:punct:]|\\s1[:punct:]|[:punct:]1[:punct:]|\\s1$|[:punct:]1$") ~ TRUE),
code_2 = case_when(str_detect(type_of_crime_discussed,"^2$|^2[:punct:]|\\s2[:punct:]|[:punct:]2[:punct:]|\\s2$|[:punct:]2$") ~ TRUE),
code_3 = case_when(str_detect(type_of_crime_discussed,"^3$|^3[:punct:]|\\s3[:punct:]|[:punct:]3[:punct:]|\\s3$|[:punct:]3$") ~ TRUE),
code_4 = case_when(str_detect(type_of_crime_discussed,"^4$|^4[:punct:]|\\s4[:punct:]|[:punct:]4[:punct:]|\\s4$|[:punct:]4$") ~ TRUE),
code_5 = case_when(str_detect(type_of_crime_discussed,"5") ~ TRUE),
code_6 = case_when(str_detect(type_of_crime_discussed,"6") ~ TRUE),
code_7 = case_when(str_detect(type_of_crime_discussed,"7") ~ TRUE),
code_8 = case_when(str_detect(type_of_crime_discussed,"8") ~ TRUE),
code_9 = case_when(str_detect(type_of_crime_discussed,"^9$|^9[:punct:]|\\s9[:punct:]|[:punct:]9[:punct:]|\\s9$|[:punct:]9$") ~ TRUE),
code_10 = case_when(str_detect(type_of_crime_discussed,"10") ~ TRUE),
code_11 = case_when(str_detect(type_of_crime_discussed,"11") ~ TRUE),
code_12 = case_when(str_detect(type_of_crime_discussed,"12") ~ TRUE),
code_13 = case_when(str_detect(type_of_crime_discussed,"13") ~ TRUE),
code_14 = case_when(str_detect(type_of_crime_discussed,"14") ~ TRUE),
code_99 = case_when(str_detect(type_of_crime_discussed,"99") ~ TRUE)
) %>%
select(state, contest, prosecutor_name, news_story_number, type_of_crime_discussed, starts_with("code")) %>%
rename(
"None" = "code_0",
"Homicide" = "code_1",
"Robbery" = "code_2",
"Rape/Sexual Assault"= "code_3",
"Arson"= "code_4",
"Burglary"= "code_5",
"Kidnapping"= "code_6",
"Theft/looting"= "code_7",
"Drunk driving"= "code_8",
"Domestic violence"= "code_9",
"Drug crimes"= "code_10",
"Shootings/gun crimes"= "code_11",
"Violent crime, broadly"= "code_12",
"Nonviolent crime, broadly"= "code_13",
"Property crime, broadly"= "code_14",
"Other"= "code_99"
) %>%
pivot_longer(!c(state, contest, prosecutor_name, news_story_number, type_of_crime_discussed),
names_to = "code",
values_to = "exists") %>%
tabyl(code,exists) %>%
select(code, `TRUE`) %>%
rename(c(crime= code, mentions = `TRUE`)) %>%
arrange(-mentions) %>%
datatable(options = list(dom='t'),
colnames = c('Crime', 'Total Mentions'),
rownames = FALSE)0 = Not mentioned
1 = Mentioned about increasing rate
2 = Mentioned about decreasing rate
3 = Mentions both increases and decreases
inc_unique_articles['discussion_of_crime_rates'][inc_unique_articles['discussion_of_crime_rates'] == "0"] <- "No Mention"
inc_unique_articles['discussion_of_crime_rates'][inc_unique_articles['discussion_of_crime_rates'] == "1"] <- "Increasing"
inc_unique_articles['discussion_of_crime_rates'][inc_unique_articles['discussion_of_crime_rates'] == "2"] <- "Decreasing"
inc_unique_articles['discussion_of_crime_rates'][inc_unique_articles['discussion_of_crime_rates'] == "3"] <- "Both"
inc_unique_articles %>%
group_by(discussion_of_crime_rates) %>%
summarize(articles=n()) %>%
mutate(pct = scales::percent(articles / sum(articles))) %>%
arrange(-articles) %>%
datatable(options = list(dom='t'),
colnames = c('Code', 'Articles', '% of Articles'),
rownames = FALSE)0 = neutral
1 = expressing concern
2 = reassuring
3 = includes both expressions of concern and reassurance
99 = not applicable
Note: Omits four articles with invalid codes for tone of coverage.
# A tibble: 4 × 23
news_story_number prosecutor_name date_of_article title_of_article
<chr> <chr> <dttm> <chr>
1 9 "Adel, Allister" 2020-01-31 00:00:00 No criminal cha…
2 72 "Adel, Allister" 2020-02-01 00:00:00 No charges for …
3 510 "John J. Flynn" 2020-02-15 00:00:00 Modify bail ref…
4 1 "Wojtaszek, Caroline \… 2020-01-05 00:00:00 Wojtaszek gets …
# ℹ 19 more variables: name_of_publication <chr>, name_of_author <chr>,
# type_of_article <chr>, duplicative_publication_of_previous_article <chr>,
# x_subj_duplicate_only <chr>, length_of_article <chr>,
# primary_focus_of_article <chr>, election_referenced_in_article <chr>,
# type_of_prosecutor_mention <chr>, type_of_crime_discussed <chr>,
# discussion_of_crime_rates <chr>,
# discussion_of_prosecutorial_decision_or_policy <chr>, …inc_unique_articles %>%
filter(tenor_tone_of_article_re_crime_or_crime_rates %in% c("0","1","2","3","99")) %>%
tabyl(discussion_of_crime_rates
,tenor_tone_of_article_re_crime_or_crime_rates)%>%
datatable(options = list(dom='t'),
colnames = c('', 'Neutral','Concern','Reassuring', 'Both','NA'),
rownames = FALSE)0 = not mentioned at all
1 = mentioned in passing
2 = mentioned with limited discussion
3 = mentioned with discussion of platforms/policies
inc_unique_articles %>%
#exclude 10 articles with invalid coding
filter(election_referenced_in_article != "99") %>%
tabyl(election_referenced_in_article) %>%
adorn_pct_formatting() %>%
datatable(options = list(dom='t'),
colnames = c('Code', 'Articles','Percent'),
rownames = FALSE)competitiveness %>%
filter(type != "open") %>%
left_join(inc_unique_articles, by = c("state", "contest")) %>%
#exclude 11 articles with invalid coding
filter(election_referenced_in_article != "99", !is.na(election_referenced_in_article)) %>%
tabyl(election_referenced_in_article) %>%
adorn_pct_formatting() %>%
datatable(options = list(dom='t'),
colnames = c('Competitiveness', 'Articles','Percent'),
rownames = FALSE)“One candidate (A. Felton, Polk County, OR) was mentioned in
eight separate articles, none of which mentioned that he was up for reelection;
he ran unopposed and won.” See Full Report
inc_unique_articles%>%
#exclude 11 articles with invalid coding
filter(election_referenced_in_article != "99", !is.na(election_referenced_in_article)) %>%
tabyl(prosecutor_name, election_referenced_in_article) %>%
datatable(options = list(dom='t',
pageLength = 27,
order = list(list(4, 'asc'),
list(3,'asc'),
list(2, 'asc'))
),
colnames = c('Prosecutor', 'No Election Mention','Passing Mention','Limited Discussion','In Depth'),
rownames = FALSE)0 = neither is mentioned
1 = case-specific/particular prosecutor decision mentioned (describe in notes)
2 = general/office-wide prosecutor policy mentioned (describe in notes)
NOTE: This table looks only at prosecutors running for re-election.
inc_unique_articles %>%
left_join(competitiveness, by=c("state", "contest")) %>%
filter(type !="open") %>%
mutate(discussion_of_prosecutorial_decision_or_policy =
case_when(discussion_of_prosecutorial_decision_or_policy == "0" ~ "neither",
str_detect(discussion_of_prosecutorial_decision_or_policy,";") ~ "both",
str_detect(discussion_of_prosecutorial_decision_or_policy,"1") ~ "decision",
str_detect(discussion_of_prosecutorial_decision_or_policy,"2") ~ "policy",
str_detect(discussion_of_prosecutorial_decision_or_policy,"General policy") ~ "policy"
)
) %>%
select(discussion_of_prosecutorial_decision_or_policy) %>%
tabyl(discussion_of_prosecutorial_decision_or_policy) %>% adorn_pct_formatting() %>%
datatable(options = list(dom='t',
order = list(2, 'desc')
),
colnames = c('', 'Articles','% Articles'),
rownames = FALSE)non_incumbents %>%
filter(duplicative_publication_of_previous_article == "0") %>%
group_by(candidate_name, candidate_background_or_platform) %>%
summarize(articles = n()) %>%
mutate(candidate_background_or_platform = case_when(
candidate_background_or_platform == "0" ~ "neither",
candidate_background_or_platform == "1" ~ "background_only",
candidate_background_or_platform == "2" ~ "platform_only",
candidate_background_or_platform == "3" ~ "both")) %>%
pivot_wider(names_from = candidate_background_or_platform, values_from = articles) %>%
replace_na(list(neither = 0,
platform_only = 0,
background_only = 0,
both = 0
)
) %>%
adorn_percentages() %>%
arrange(desc(neither)) %>%
adorn_pct_formatting() %>%
datatable(
options(list(dom='t',
pageLength = 21)),
colnames = c('Candidate',
'Neither',
'Platform',
'Background',
'Both'),
rownames = FALSE)