Did research institutions sponsor hybrid open access journals that currently share open license metadata with Crossref?

cr_opac.R

library(jsonlite)
library(tidyverse)
library(ggalt)
#' get data from the open apc initiative
u <-
  "https://raw.githubusercontent.com/OpenAPC/openapc-de/ec458eaf0232eb193bb5f8e3b8481491e39a60ef/data/apc_de.csv"
o_apc <- readr::read_csv(u) %>%
  filter(is_hybrid == TRUE)
#' Some summary statistics:
#'
#' We also would like to add data from offsetting aggrements, which is also collected
#' Open APC initiative, but does not include pricing information.
#' 
o_offset <- readr::read_csv("https://raw.githubusercontent.com/OpenAPC/openapc-de/ec458eaf0232eb193bb5f8e3b8481491e39a60ef/data/offsetting/offsetting.csv")
#' Merge with Open APC dataset
o_apc <- o_offset %>% 
  mutate(euro = as.integer(euro)) %>% 
  bind_rows(o_apc) 
#' Include country information, which are available via Open APC OLAP server: 
#' <https://github.com/OpenAPC/openapc-olap/blob/master/static/institutions.csv>
country_apc <- readr::read_csv("https://raw.githubusercontent.com/OpenAPC/openapc-olap/master/static/institutions.csv") %>%
  select(institution, country)
countries <- readr::read_csv("https://raw.githubusercontent.com/OpenAPC/openapc-olap/master/static/institutions_offsetting.csv") %>%
  bind_rows(country_apc) %>%
  distinct() %>% 
  mutate(country = gsub("NDL", "NLD", country)) %>%
  mutate(country_name = countrycode::countrycode(country, "iso3c", "country.name"))
#' merge with open apc dataset
#' 
o_apc <- o_apc %>%
  left_join(countries, by = "institution")
#' main indicator file from hybrid oa dashboard
jns <- jsonlite::stream_in(file("https://raw.githubusercontent.com/subugoe/hybrid_oa_dashboard/master/data/jn_facets_df.json")) %>%
  distinct(issn, journal_title, publisher)

o_apc_country <- o_apc %>%
  left_join(jns, by = "issn") %>%
  group_by(country, journal_title, period) %>%
  summarize(oapc_n_country = n()) %>%
  filter(period %in% c("2013", "2014", "2015","2016", "2017")) %>%
  ungroup() %>%
  group_by(country, journal_title) %>%
  summarize(oapc_total = sum(oapc_n_country))
#' data from hybrid oa dashboard
o_apc_ind <- jsonlite::stream_in(file("https://raw.githubusercontent.com/subugoe/hybrid_oa_dashboard/master/data/oapc_aggregated.json")) %>%
  group_by(country, journal_title) %>%
  summarize(oapc_cr = sum(oapc_n_country))

my_df <- left_join(o_apc_country, o_apc_ind, by = c("country", "journal_title"))

cr_jns <- my_df %>% 
  filter(!is.na(oapc_cr)) %>% 
  group_by(country) %>% 
  summarise(cr_jns = n_distinct(journal_title))

jns <- my_df %>% 
  group_by(country) %>% 
  summarise(jns = n_distinct(journal_title)) %>%
  inner_join(cr_jns, by = ("country")) %>% 
  arrange(desc(jns)) %>%
  mutate(country = forcats::as_factor(country))

gg <- ggplot(jns,
       aes(x = jns, xend = cr_jns, y = country, color = group)) +
  ggalt::geom_dumbbell(
    colour="#30638E",
    colour_xend="#EDAE49",
    colour_x="#30638E",
    size_x=3.5,
    alpha = 0.9,
    size_xend = 3.5
  ) +
  scale_y_discrete(limits = rev(levels(jns$country))) +
   scale_x_continuous(breaks = seq(0, 3500, by = 500)) +
  labs(x = "Number of Hybrid OA Journals", 
       y = NULL,
       title = "Did research institutions sponsor hybrid open access journals\nthat currently share open license metadata with Crossref?",
       subtitle = "Notice that only those hybrid open access journals were included where\nacademic institutions sponsored the publication fee according to the Open APC initiative"
  ) +
  geom_text(data=jns, 
            aes(x=2300, y= "GBR", label="with Crossref licensing infos"),
            color="#EDAE49", hjust=1, size=3, nudge_x=-10) +
  geom_text(data=jns, 
            aes(x=3500, y= "GBR", label="All"),
            color="#30638E", hjust=0, size=3, nudge_x=10) +
  theme_minimal(base_family="Arial Narrow") +
  theme(plot.margin=margin(30,30,30,30)) +
  theme(panel.grid.minor=element_blank()) +
  theme(axis.ticks=element_blank()) +
  theme(panel.grid.major.y=element_blank()) +
  theme(panel.border=element_blank())
ggsave(gg, filename = "licensing_coverage_country.png", width = 9, height = 6, dpi = 450, device = "png")