Frontier Set

---

Here is some R experimentation using IPEDS data for a group of institutions called the Frontier Set. For this exercise I deliberately choose variables that had 2016 information available for each Frontier Set institution. I deleted two cases (institutions) that didn’t have any data on student graduation rates.

Variables in Dataset

Graduation Rate: IPEDS graduation rates (GR) include only full-time, first-time, degree-/certificate-seeking students who started and finished at the same institution. GRs do not represent all of the students at an institution (e.g., GR excludes part-time and transfer students). The GRs in this dataset are calculated as the # of students who completed their program within 150% of normal time divided by the # of students in the entering cohort. Student subgroups include African American, Hispanic, African American + Hispanic, and White. If you’re interested you can read more about IPEDS Graduation Rates.

Services: These are services/supports that may or may not be provided by an institution. They include: weekend/evening courses, remedial services, academic/career counseling, employment services, placement services for completers, and on-campus daycare for students’ children.

Degree of Urbanization: Represents the urbanicity by population size of the institution’s location. The coding is based on a methodology developed by the U.S. Census Bureau’s Population Division in 2005. In this dataset I recoded urbanization into a single locale (e.g., the codes ‘City: Large’, ‘City: Midsize’, and ‘City: Small’ are combined into the single code ‘City’).

Full- and Part-Time Instructional Staff: An occupational category that includes staff who are either: 1) Primarily Instruction, or 2) Instruction combined with research and/or public service. IPEDS’s intent for this category is to include all individuals whose primary occupation includes instruction at the institution.

---

Here’s the data frame—not too exciting.

library(tidyverse)
fset <- read.csv("ipeds4.27_v5.csv", header = TRUE)
fset

---

Below are some descriptives. The Frontier Set has almost the same number of 2- and 4-year institutions. Over half of the 2-institutions are campuses of the Tennessee College of Applied Technology. (The Tennessee Board of Regents is part of the Frontier Set.)

Let’s look at urbanization: Most of 4-year institutions are in cities, followed by towns, suburbs, and rural areas. The 2-year institutions have a much more even spread.

ggplot(data = fset) + 
  geom_bar(mapping = aes(x = IHE.Level, fill = Urban))

---

The plot below shows the maximum, minimum, and median values for minority student graduation rates by urbanicity.

ggplot(data = fset) + 
  stat_summary(
    mapping = aes(x = Urban, y = GR.AfAmHis),
    fun.ymin = min,
    fun.ymax = max,
    fun.y = median
  )

Same information below but now in a boxplot, with color representing the type of institution. You’ll notice there’s one suburban 2-year institution that’s an outlier. I went back to the data to investigate and it’s the Tennessee College of Applied Technology in Jackson, TN. Their GR for minority students is 69% (for white students it’s 83%). The institution with the next highest GR rate for minority students (18%) that’s also in the suburbs is Volunteer State Community College (their GR for white students is only 26%). This time I went the extra mile and added labels to the plot :)

ggplot(fset, aes(Urban, GR.AfAmHis)) +
   geom_boxplot(aes(color = IHE.Level)) +
     labs(title = "Minority student GRs by Location and Level of IHE") +
     labs(
      x = "Location of Institutions",
      y = "GR of Minority Students",
    color = "Institution Type"
)

---

Moving on, here’s the relationship between minority student graduation rates and the number of full-time instructional staff.

ggplot(data = fset) + 
  geom_point(mapping = aes(x = InstFT , y = GR.AfAmHis, color = IHE.Level)) 

ggplot(fset, aes(InstFT, GR.AfAmHis)) +
  geom_point(aes(color = IHE.Level)) +
  labs(title = "GRs of minority students by FT instructional staff") +
  labs(
    x = "Number of full-time instructional staff",
    y = "GRs of Minority Students",
    color = "Institution Type"
  )

The plot below has the same information but I added the shape aesthetic to identify institutions that offer weekend and evening courses.

ggplot(data = fset) + 
  geom_point(mapping = aes(x = InstFT, y = GR.AfAmHis, color = IHE.Level, shape = wnd.eve))

ggplot(fset, aes(InstFT, GR.AfAmHis)) +
  geom_point(aes(color = IHE.Level, shape = wnd.eve)) +
   labs(title = "GRs of minority students by FT instructional staff") +
   labs(
    x = "Number of full-time instructional staff",
    y = "GRs of Minority Studentss",
    color = "Institution Type",
    shape = "IHE offers weekend/evening courses"
  )

---

That’s the end of my experiment. Some thoughts:

• No surprise, but I found it much more fun to work with a dataset that I care about.
• I ended up doing most of the data transformation in Excel (e.g., recoding, renaming, and creating derived variables). I got hung up trying to do this in R for over a week, so for now I’ll take the path of least resistance and do what works so I can keep moving.
• The spell-check feature in R is awesome—I can’t tell you how many times I misspelled “institution.”