Privatization of AI research: Descriptivess

Load data

data <- read_rds('../../temp/data_reg.rds')

In between: Check dropouts

data_dropout <- data %>%
  group_by(AID) %>%
  mutate(dropout = year == max(year, na.rm = TRUE)) %>%
  ungroup() %>%
  filter(year <= 2019) %>%
  mutate(seniority_bin = case_when(
      seniority <= 5 ~ "1 early",
      seniority > 5 & seniority <= 10 ~ "2 mid",
      seniority > 10 & seniority <= 15 ~ "3 mid late",
      seniority > 15 ~ "4 late",
  )
  )

data_dropout %>% 
  filter(year >= 2015, year <2020) %>%
  count(year, dropout) %>%
  group_by(year) %>%
  mutate(n = n / sum(n)) %>%
  ungroup() %>%
  mutate(year = year %>% as.character()) %>%
  filter(dropout == TRUE) %>%
  ggplot(aes(x = year, y = n)) +
  geom_col()

n_dropout <- data_dropout %>%
  filter(dropout == TRUE, type == 'switcher') %>%
  count(year, seniority_bin) %>% 
  group_by(year) %>%
  mutate(pct = n / sum(n)) %>%
  ungroup()

n_dropout %>%
  mutate(year = year %>% as.character()) %>%
  ggplot(aes(x = year, y = n, fill = seniority_bin, col = seniority_bin)) +
  geom_col() +
  labs(title = 'Dropouts total')

n_dropout %>%
  mutate(year = year %>% as.character()) %>%
  ggplot(aes(x = year, y = pct, fill = seniority_bin, col = seniority_bin)) +
  geom_col() +
  labs(title = 'Dropouts total')