Skip to contents

Following the template in OpenAlex’s oa-percentage tutorial, this vignette uses openalexR to answer:

How many of recent journal articles from the University of Pennsylvania are open access? And how many aren’t?

We first need to find the openalex.id for University of Pennsylvania. We can do this by fetching for the institutions entity and put “University of Pennsylvania” in display_name or display_name.search:

oa_fetch(
  entity = "inst", # same as "institutions"
  display_name.search = "\"University of Pennsylvania\""
) %>%
  select(display_name, ror) %>% 
  knitr::kable()
display_name ror
University of Pennsylvania https://ror.org/00b30xv10
California University of Pennsylvania https://ror.org/01spssf70
Hospital of the University of Pennsylvania https://ror.org/02917wp91
University of Pennsylvania Health System https://ror.org/04h81rw26
Indiana University of Pennsylvania https://ror.org/0511cmw96
Cheyney University of Pennsylvania https://ror.org/02nckwn80
University of Pennsylvania Museum of Archaeology and Anthropology https://ror.org/02aej5879
Raymond and Ruth Perelman School of Medicine at the University of Pennsylvania https://ror.org/041phzg73

We will use the first ror, 00b30xv10, as one of the filters for our query.

Alternatively, we could go to the autocomplete endpoint at https://explore.openalex.org/ to search for “University of Pennsylvania” and find the ror there!

All other filters are straightforward and explained in detailed in the original jupyter notebook tutorial. The only difference here is that, instead of grouping by is_oa, we’re interested in the “trend” over the years, so we’re going to group by publication_year, and perform the query twice, one for is_oa = "true" and one for is_oa = "false" .

open_access <- oa_fetch(
  entity = "works",
  institutions.ror = "00b30xv10",
  type = "article",
  from_publication_date = "2012-08-24",
  is_paratext = "false",
  is_oa = "true",
  group_by = "publication_year",
  count_only = TRUE
)

closed_access <- oa_fetch(
  entity = "works",
  institutions.ror = "00b30xv10",
  type = "article",
  from_publication_date = "2012-08-24",
  is_paratext = "false",
  is_oa = "false",
  group_by = "publication_year",
  count_only = TRUE
)

uf_df <- closed_access %>%
  select(- key_display_name) %>%
  full_join(open_access, by = "key", suffix = c("_ca", "_oa")) 

uf_df
#>     key count_ca key_display_name count_oa
#> 1  2023     5978             2023     6738
#> 2  2020     5198             2020    10179
#> 3  2021     4855             2021    10315
#> 4  2022     4855             2022     9015
#> 5  2019     4822             2019     8292
#> 6  2018     4786             2018     7756
#> 7  2015     4543             2015     5757
#> 8  2016     4497             2016     6261
#> 9  2014     4371             2014     5447
#> 10 2017     4187             2017     6956
#> 11 2013     4128             2013     5195
#> 12 2012     1148             2012     1433
#> 13 2024       49             2024        6

Finally, we compare the number of open vs. closed access articles over the years:

uf_df %>%
  filter(key <= 2021) %>% # we do not yet have complete data for 2022 and after
  pivot_longer(cols = starts_with("count")) %>%
  mutate(
    year = as.integer(key),
    is_oa = recode(
      name,
      "count_ca" = "Closed Access",
      "count_oa" = "Open Access"
    ),
    label = if_else(key < 2021, NA_character_, is_oa)
  ) %>% 
  select(year, value, is_oa, label) %>%
  ggplot(aes(x = year, y = value, group = is_oa, color = is_oa)) +
  geom_line(size = 1) +
  labs(
    title = "University of Pennsylvania's progress towards Open Access",
    x = NULL, y = "Number of journal articles") +
  scale_color_brewer(palette = "Dark2", direction = -1) +
  scale_x_continuous(breaks = seq(2010, 2024, 2)) +
  geom_text(aes(label = label), nudge_x = 0.1, hjust = 0) +
  coord_cartesian(xlim = c(NA, 2022.5)) +
  guides(color = "none")