The chief administrator job of a higher education institution is, as the title implies, the pinnacle of careers in academic administration. The chief administrator is a school’s spokesperson and guides its vision, affecting the lives of the thousands of students who pass through those institutions. And for the chief administrators who don’t care about the idealism of educating future generations, I would imagine the high-six/low-seven figure salaries that many earn is incentive enough. Despite the prominent roles that these administrators fill, there is a dearth of publicly available data on them.

Every institution that participates in federal student financial aid is required to submit data to Integrated Postsecondary Education System, or IPEDS. Eventually, the submitted data is made publicly available in the IPEDS Data Center. This includes numbers on admissions, student enrollment, degree completions, graduation rates, financial aid, finances, human resources, and libraries. Much of this data has to be reported by gender and race/ethnicity. For example, how many Hispanic female undergraduates began at an institution last fall? What is the graduation rate of male American Indian Alaska Natives? What is the average salary of female instructional staff on a 9-month contract? But for one reason or another, as far as I can tell, the only things that must be reported about an institution’s chief administrator are her or his name and title.

Names, though, are not completely devoid of meaningful information. If you live in the United States and hear the name Steven, you probably think of a male, and if you hear the name Mary, you probably think of a female. Yes, some names are more ambiguous than others (my own being a good example), and some people’s names might belie the gender they identify with, but there is a degree of reliability that a person’s name offers in determining whether they are female or male. Thus, using first names to make educated guesses about chief administrators’ gender, my goal here was to describe gender representation among this set of individuals.

The first step was to download the names of the chief administrators for every institution in the IPEDS Data Center (n = 7108) and clean up the data. This required putting years in a consistent format; removing titles preceding first names (e.g., Dr., Ms., Mrs.); extracting first names into their own column; and adding variable labels for college sector.

library(tidyverse)
library(babynames)
library(scales)

admin <- read_csv("data/chief-admin-names.csv")

# tidy year 
admin <- admin %>% 
  gather("year", "name", `Name of chief administrator (HD2016)`,
         `Name of chief administrator (HD2015)`:`Name of Chief Administrator (IC90HD)`) 

# extract year from names and put in consistent format
admin <- admin %>% 
  mutate(year =  parse_number(year),
         year = ifelse(year >= 9596 & year <= 9798,
                       str_sub(start = 1, end = 2, year), year),
         year = ifelse(nchar(year) == 2, 
                       paste0("19", year), 
                       year),
         year = as.integer(year))

# select and rename variables
admin <- admin %>% 
  select(unit_id = UnitID, 
         sector_code = `Sector of institution (HD2016)`,
         institution_name = `Institution Name`,
         year, name,
         undergrad_enroll_2016 = `Grand total (EF2016  All students  Undergraduate total)`,
         grad_enroll_2016 = `Grand total (EF2016  All students  Graduate and First professional)`)

# titles are in the first position for many names, so need to remove
# those so can extract first posistion from names and have it reflect
# first name
admin <- admin %>% 
  mutate(name = tolower(name),
         name = gsub("\\.", "", name),
         name = gsub("\\<dr\\>", "", name),
         name = gsub("\\<mr\\>", "", name),
         name = gsub("\\<ms\\>", "", name),
         name = gsub("\\<mrs\\>", "", name),
         name = gsub("\\<rev\\>", "", name),
         name = gsub("\\<reverend\\>", "", name),
         name = gsub("\\<very reverend\\>", "", name),
         name = gsub("\\<very\\>", "", name),
         name = gsub("\\<rabbi\\>", "", name),
         name = gsub("\\<msgr\\>", "", name),
         name = gsub("\\<dra\\>", "", name),
         name = gsub("\\<sr\\>", "", name),
         name = str_trim(name, side = "both"),
         full_name = tolower(name)) %>% 
  separate(name, into = "first_name", sep = " ")

# add sector label
labels <- read_csv("data/sector-value-labels.csv") %>% 
  select(sector_code = Value, 
         sector_label = ValueLabel)

admin <- left_join(admin, labels, by = "sector_code")

Next, I assigned a gender to each chief administrator based on her or his first name. To do this, I used R’s babynames package, which contains the number of babies born every year from 1880 to 2015 for each combination of name and sex¹. I took the babynames data and calculated how often each name was given to females vs. males, and then assigned gender based on the higher proportion. For example, in the babynames data, about 72% of all newborns named Jaydin were male, so I assigned the name Jaydin to male. Names like Jaydin, however, were the exception: Most of the time, names went overwhelmingly to one sex or the other, with the vast majority of names in the babynames dataset being associated with only one sex.

# read in babynames data and assign a proportion to female and male for each name
baby_names <- babynames %>% 
  select(sex, name, n) %>% 
  mutate(name = tolower(name))

baby_names <- baby_names %>% 
  group_by(sex, name) %>% 
  summarise(total = sum(n)) %>% 
  ungroup() %>% 
  group_by(name) %>% 
  mutate(prop = total/sum(total)) %>%
  filter(prop == max(prop)) %>% 
  ungroup() %>% 
  select(sex, first_name = name, prop)

# one chief admin officer has name gold and it happens to be one that is .5 prop,
# so removed from dataset. (gold was given as full name, so i'm *guessing* this 
# is actually last name).
baby_names <- baby_names %>% 
  filter(first_name != "gold" | prop != .5)

# join ipeds and babynames, removing rows where there were no matches
admin <- left_join(admin, baby_names, by = "first_name") %>% 
  mutate(institution_name = gsub("-", " ", institution_name)) %>% 
  filter(!is.na(sex))

admin <- admin %>% 
  select(unit_id, institution_name, sector_label, year, undergrad_enroll_2016, 
         grad_enroll_2016, full_name, first_name, sex)

# several variables are for 2016 only, so put those in their own data frame
admin_2016 <- admin %>% 
  filter(year == 2016) %>% 
  select(-year)

admin <- admin %>% 
  select(-undergrad_enroll_2016, -grad_enroll_2016)

# proportion female by year
female_prop_sex <- admin %>% 
  count(year, sex) %>%
  group_by(year) %>% 
  mutate(year_total = sum(n)) %>% 
  ungroup() %>% 
  mutate(prop_sex = n/year_total) %>% 
  filter(sex == "F")
  
female_prop_sex %>% 
  ggplot(aes(x = factor(year), y = prop_sex, group = 1)) +
  geom_line() +
  geom_point() +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1)) +
  scale_y_continuous(label = percent_format()) +
  labs(x = "Year", y = "Percent Women",
       title = "Percentage of Chief Administrators Who Are Women",
       subtitle = "All IPEDS Institutions, Years 1990-2016")

# proportion female by sector
female_by_sector <- admin_2016 %>% 
  count(sector_label, sex) %>% 
  group_by(sector_label) %>% 
  mutate(prop_sex = n/sum(n)) %>% 
  ungroup() %>% 
  filter(sex == "F")

female_by_sector %>%
  ggplot(aes(x = reorder(sector_label, prop_sex), y = prop_sex)) +
  geom_col() +
  coord_flip() +
  scale_y_continuous(label = percent_format()) +
  theme_minimal() +
  labs(x = NULL, y = "Percent Women",
       title = "Percentage of Chief Aministrators Who Are Women,\nby Sector",
       subtitle = "All IPEDS Institutions, Year 2016")