Sign in

International students in the U.S: What the data tells us

Background

The United States has historically been one of the most popular destinations for international students around the world, as educational institutions in the U.S offer valuable professional and academic opportunities. According to the president of Pace University, colleges also benefit from hosting international students. Not only do they help create unique cross-cultural experiences in classrooms, international students are becoming “increasingly important to keep our classes full, our tuition revenue up, and our institutions thriving.”

Project Motivation

As an international student myself, I have been fortunate enough to study at two amazing schools in the U.S in the past 8 years (Holderness School and the University of Pennsylvania.) Recently, I came across an article detailing an unprecedented decrease in the number of international students in the U.S in the 2019–2020 school year. The COVID-19 pandemic, the U.S government’s “lack of a coordinated national pandemic response”, and widespread uncertainty over travel restrictions led to a sharp decline in enrollments of new international students from the prior school year.

Using R to analyze datasets from organizations like the Institute of International Education, I sought to gain more detailed insights into international students in U.S colleges. Trends in international student enrollments can offer a glimpse into the state of higher education in the U.S as well as how other countries’ perception of U.S-based educational institutions (and the U.S as a whole) has changed over the years. It can also help universities make more informed decisions about international student recruitment and retention.

I will first analyze historical trends of international students in the U.S. Then I will present a breakdown of international students in the U.S by key features, such as country/continent of origin, where they study in the U.S, and field of study. Next, I will delve deeper into U.S colleges and their international student population. Finally, I will run a regression analysis to understand whether certain geographical and college-specific factors influence the number of international students in different U.S colleges.

Data Sources

  1. Institute of International Education: Opendoors Data

Important Note: “Year” = School year. For example, “2019” refers to the 2019–2020 school year that began in fall 2019.

Analysis

I. Overview

How has the number of international students in the U.S changed over time?

This graph depicts trends in both the total number of international students (bar chart) and international students as a percentage of total enrolled students (line chart) from 1969 to 2019.

The number of international students in the U.S has increased rapidly since 1969, surpassing 1 million students for the first time in the 2015–2016 school year. While international students represented less than 2% of total students in the U.S in the 1969–70 school year, that number rose to over 5% by 2019–20. We can (sort of) see the decline in international students in recent years, but observing the annual percent change will make it more clear.

In the last 70 years, there was only a handful of times when the number of international students in the U.S decreased: the 1971–72 school year, between 2003–04 and 2005–06 (potentially due to effects of 9/11), and finally, the most recent 2019–20 school year. Given that the number of international students has increased consistently since 2005, even during the Great Recession in 2008, it makes sense why experts have called the most recent decline “unprecedented.”

How do historical trends vary by academic level?

The number of undergraduate, graduate, non-degree students have been declining in recent years, while students on OPT (temporary work visa offered to international students) have increased rapidly in the past 10 years. There are currently more undergraduate international students than graduate international students in the U.S, but from around 2000 to 2010, there were more graduate students than undergraduates.

Some experts argue that observing changes in enrollments of new international students is a more important than measuring the number of total international students. Therefore, I will now look at changes in enrollments of new international students from 2008 to present.

While the past school year’s decline in new international students was largely driven by the COVID-19 pandemic’s impact, it appears that the number of newly enrolled international students has actually been dropping over the past few years, starting in the 2016–17 school year. The decline was the sharpest in the 2017–18 school year.

I hoped to better understand the cause of this decline and came across a recent survey of international students that found the US’s visa process, travel restrictions, high tuition costs, as well as an “unwelcoming” political climate shaped by the the Trump administration’s anti-immigration stance and xenophobic rhetoric to be the main reasons why less and less international students have enrolled in U.S universities in the past years.

The same article pointed out that countries like the U.K., Canada and Australia have seen spikes in international student enrollments in the last few years, which suggests that the cause behind the decline may potentially be specific to the U.S.

I personally would like to read more about how past presidents’ stance on immigration and the legislative changes each administration implemented have (or have not) affected new enrollments of international students in the U.S in the past. I’m curious to see if the recent downward trend in international student enrollment will be reversed under the new administration, given that the Biden administration has proposed to make broad changes to US immigration policies (that may potentially be more favorable to international students) and used a more immigrant-friendly rhetoric during the presidential campaign.

How do new enrollment trends vary by academic level?

The number of new international students pursuing non-degrees, which include foreign students pursuing short-term exchanges and intensive English language programs, has decreased by 28.4% since the the 2014–15 school year. New international undergraduate and graduate students have both been in decline since the 2016–17 school year. Yet, it appears that the recent drop in new international students is driven more by the reduction of new undergraduate international students than international graduate students. While new enrollments of international graduate students have began to rise again in the last two years, the number of new international undergraduate students has continued to decline.

II. Breakdown of international students in the U.S by key features

Where are most international students located within the U.S?

The highest proportion of international students are located in the Northeast. Compared to 2014, a higher percentage of international students are located in the South and the Northeast in 2021, while the share of international students living in the Midwest and the West declined. More research is needed to determine the exact reason why the share of students studying in the Northeast remains the largest and continues to grow. However, it may be partially due to the fact that the American domestic college-age population in the Northeast is shrinking more than other regions, leaving colleges in the Northeast to rely more on the enrollment of international students.

Naturally, states with a large population (CA, NY, TX, FL) appeared to have the highest number of international students, followed by northeastern states like PA and CT. Therefore, comparing the international student population in each state per capita would be more reasonable.

Which U.S states have more international students (per capita) than others?

It appears that states in the Northeast region (+ California) have the largest international student population per capita. Kentucky is the only Southern state included in this list. It is worth noting that the Department of Homeland Security’s data, which was used for this analysis, also included international students in high schools, unlike the IIE data used in the analysis above, which only counted students in U.S. colleges or universities.

These states in the South and the Midwest have the smallest international student population per capita. I would love to learn more about the factors that drive state-by-state differences in states’ international student population. For example, I’m curious why Kentucky has a much higher international students population per capita than its neighboring states, Tennessee and Arkansas.

Which continents do international students come from, and how has that changed over the years?

The share of U.S international students coming from Asia rose dramatically from 55.14% in the 2000–01 school year to 70.48% in the 2019–20 school year. The share of U.S international students coming from the rest of the continents declined, with the exception of Middle East & North Africa.

Which countries do international students come from, and how has that changed over the years?

Together, Chinese and Indian students comprised 52.5% of international students in the 2019–20 school year, with Chinese students accounting for 34.6% of international students, which is almost twice as much as Indian students do. Has China always been the largest single source of international students in U.S universities? What do historical trends tell us?

To find out, I first selected the top countries international students hail from by computing the historical average of the number of international students in the U.S by country of origin from 2000 to 2019. I learned that the Top 6 countries of origin were: China, India, South Korea, Japan, Canada, and Saudi Arabia.

Interestingly, China hasn’t always been the largest source of international students in U.S universities. India actually sent more international students to the U.S than China until about 2008, when the number of Chinese international students in the U.S started growing very rapidly, outpacing the rest of the countries.

In contrast, while Japan was the #3 largest source of international students in the U.S in 2000, the number of Japanese students declined steadily throughout the last 20 years. Japan and South Korea sent similar number of students to the U.S in 2000, but the number of Korean students soon surpassed Japanese students. Yet, the number of Korean students also began to steadily decline starting in 2008. In fact, with the exception of India and China, the number of international students coming from the rest of the top countries has either stalled or decreased since 2000.

The number of Indian international students in the U.S declined in the 2019–20 school year, which may be temporary, as it is likely due to effects of the COVID-19 pandemic. The number of Chinese international students increased during the 2019–20 school year, even in the face of the COVID-19 pandemic. Given the unparalleled, rapid growth in the number of Chinese international students in the U.S, I wonder if the number of Chinese student studying in the U.S will continue to rise rapidly in the next years as well, in spite of recent tensions between the two countries.

Examining the recent decline in the number of international students

Given that the drop in the number of international students in the 2019–20 school year was “unprecedented,” I wanted to know whether the decline was more severe for students from certain countries or regions.

To find out, I computed the percent change of the number of international students between the 2018–19 and the 2019–20 school years, both by country and by region.

In general, it seemed like the decline was more rapid for countries in the Middle East, with Saudi Arabia experiencing the sharpest decline in the number of international students in the U.S. Countries in Central America and Central/Southeast Asia also seemed to have experienced large drops. While useful, these graphs must be interpreted with caution, as some countries and regions, such as Southern and North Africa, didn’t send that many international students to begin with.

Which academic subjects are most popular among international students in the U.S? How has that changed over the years?

While ‘Business & Management’ has been the #1 most popular field of study among international students in the past, ‘Engineering’ has overtaken ‘Business & Management’ as the #1 most popular subject around 2014, and ‘Math & Computer Science’ rose to #2 around 2017. It appears that the number of international students pursuing business has been notably declining in recent years, while STEM-based fields of studies are rising in popularity. This may be linked to the fact that companies in the U.S are increasingly prefer to recruit international students with STEM backgrounds. The fact that students who pursue STEM subjects are eligible to work in the U.S for 3 years following their graduation under the OPT program, unlike the usual 1 year for students who pursue non-STEM majors, may also be contributing to the rising popularity of STEM fields among international students.

III. International student population in colleges: Exploratory Analysis

Which U.S colleges have the largest international student population?

In the 2019–20 school year, New York University, Northeastern University, University of Southern California, and Columbia University hosted the most international students. Noticing that many of these colleges are based in cities with high population density (NYC, LA, BOS,) I wanted to delve deeper into whether certain geographical (i.e. population density) and college-specific variables (tuition, acceptance rate, endowment, etc.) are related to the size of colleges’ international student population. Because large universities may naturally have a larger international student population, international students as a % of the total student body will be my main variable of interest from here on.

How do Ivy league universities compare with each other in terms of their international student population?

Looking into differences across Ivy League universities seemed interesting, as these schools are relatively comparable in terms of selectiveness and academic quality. In the graph below, I analyzed international students as a share of the student body for all 7 Ivy League universities.

Columbia university (NYC) and UPenn (Philadelphia) had the highest percentage of international students, while Dartmouth College (Hanover, NH) had the lowest percentage of international students. At first glance, the results seemed to validate my initial hunch about the relationship between colleges’ international student population and the population density of the city they are located in. My next step was to run a regression analysis, with international students as a share of all students as the outcome variable.

IV. International student population in colleges: Regression Analysis

Do certain geographical and college-specific factors influence the number of international students in different U.S colleges?

In order to run the regression analysis, I scraped colleges’ international student stats from collegedata.com, detailed information about each college from U.S News, and ZIP-code level geographical data from a R package called ‘zipcodeR.’ Using Tableau Prep and R, I cleaned and merged the three datasets into one that ultimately contained 477 observations (= colleges.)

Out of all the colleges in my dataset, the universities listed above have the highest percentage of international students as a share of all students. Interestingly, all 10 schools in this list were art schools in large cities like New York, Chicago, and San Francisco.

When choosing which information to scrape, I mainly relied on my intuition to derive hypotheses about the relationship between that particular variable and the percentage of international student in each college, as outlined in the chart below. For instance, I predicted that schools’ total tuition and the percentage of international students will have a positive relationship, as I assumed colleges that charge higher out-of-state tuition will have a higher percentage of international students. Read more below.

Before going ahead and using all of the predictor variables above, I first had to check for multicollinearity.

Perhaps unsurprisingly, ‘median home value’ and ‘median household income’ appeared to be highly correlated. Because ‘median home value’ is also highly correlated with ‘population density’, I decided not to use ‘median home value’ as one of my predictor variables. Additionally, ‘grad_rate’ appeared to be correlated with both ‘acceptance_rate’ and ‘faculty_student_ratio’, so I decided not to select ‘grad_rate’ as well.

Preliminary Results

After standardizing the numeric variables, I fitted a multiple regression model, with ‘perc_intl_students’ as the outcome variable and all variables except median home value and graduation rate as predictor variables.

Because ‘X2019_endowment’ and ‘starting_salary ’ did not appear to be statistically significant variables, I decided not to include them in my final model.

Final Results and Interpretation

By performing a regression analysis on this dataset, I found that most variables included were statistically significant, though some (‘faculty_student_ratio’, ‘total_enrollment’, and ‘median_household_income’) were less statistically significant than others (‘class’, ‘total_tuition’, and ‘acceptance_rate.’) Because all numerical variables were scaled, standardized coefficients estimates can be interpreted as “the number of standard deviation units Y (perc_intl_students) changes with an increase in one standard deviation in X.” For example, the coefficient for ‘total_tuition’ is 1.36, meaning that an increase in one standard deviation in ‘total_tuition’ leads to a 1.36% increase in the percentage of international students.

Compared to art schools (which was the baseline category for the ‘class’ variable,) liberal arts schools, private universities, and public universities all seem to have a lower percentage of international students on average.

In terms of directionality, the sign of the correlation coefficients seemed to validate most of my initial hypotheses. ‘total_tuition’ and ‘perc_intl_students’ had a positive relationship, while ‘acceptance_rate’ and ‘perc_intl_students’ had a negative relationship, both of which matched my earlier predictions. Similarly, ‘population_density’ and ‘perc_intl_students’ also had a positive relationship, suggesting that colleges in areas with high population density, in fact, do have a higher percentage of international students on average.

‘Total_enrollment’, ‘faculty_student_ratio,’ and ‘median_household_income’ all had positive relationships with ‘perc_intl_students’, but again, the predictor variables were not as statistically significant as the rest.

Because all numerical variables were standardized, I was able to compare the variable importance of each predictor variable by comparing their coefficient estimates. Changes in the standard deviation of ‘Total_tuition’ will lead to the greatest change in colleges’ percentage of international students, while changes in the standard deviation of ‘median_household_income’ will lead to the smallest change.