Thursday, May 10, 2018

Looking at Transfers

It's official: Princeton has broken its streak of not considering transfer students for admission, and has admitted 13 applicants for the Fall, 2018 term of the 1,429 who applied, for an astonishing how-low-can-you-go admit rate of 0.9%.  Of course, we'll have to wait until sometime in the future to see how many--if any--of them actually enroll.

I thought it might be interesting to take a look at transfers, so I did just that, using an IPEDS file I had on my desktop.  There are four views here, and they're pretty straightforward:

The first tab shows the number of transfers enrolled by institution in Fall, 2016 (left hand column) and the transfer ratio.  The ratio simply indicates how many new transfer students for Fall, 2016 you'd meet if you were to go on that college campus in Fall, 2016 and choose 100 students at random.  A higher number suggests a relatively more transfer friendly institution. You can choose any combination of region, control and broad Carnegie type using the filters at the top.

The second tab shows the same data arrayed on a scatter gram; type any part of a college name and then select it to see it highlighted on the chart.  Hover over a point for details.

The third chart is static, and shows undergraduate enrollment in Fall, 2016 and the number of new transfer students in the same term.  The bars are split by region and colored by Carnegie type.

And the last tab shows the weighted transfer ratios, split the same way.

As you'll see, thirteen students doesn't seem so significant against the 810,000 new transfers in Fall, 2016.  But it's a start.




Monday, May 7, 2018

Want to increase graduation rates? Enroll more students from wealthier families.

OK. Maybe the headline is misleading.  A bit.

I've written about this before: The interconnectedness of indicators for colleges success.  This is more of the same with fresher data to see if anything has changed. Spoiler alert: Not much.

What's new this time is the IPEDS publication of graduation rates for students who receive Pell and those who don't, along with overall graduation rates.  While the data are useful in aggregate to point out the trends, at the institutional level, they are not.

First, some points about the data:  I've included here colleges with at least 20 Pell-eligible freshmen in 2015, just to eliminate a lot of noise.  Colleges with small enrollments don't always have the IR staff to deliver the best data to IPEDS, and they make the reports a bit odd.  And even without these institutions, you see some issues.

Second, colleges that do not require tests for admission are not allowed to report tests in IPEDS.  Once you check "not required" that box with test scores gets grayed out, so attempting to report them is futile.

But, it's here.  View one shows pretty much every four-year public and private not-for-profit college in the US, and includes four points: On the left as dots are six-year grad rates for all students (light blue), Pell students, (dark blue) and all students (purple).  On the right is the gap between Pell grad rates and non-Pell students.  Again, some of these numbers are clearly wrong, or skewed by small numbers in spite of the exclusion noted above.

The next four collectively tell the story of wealth and access:


  • If you have more Pell students, your graduation rate is lower
  • While most colleges do a pretty good job of keeping Pell and non-Pell grad rates close, there are some troubling outliers
  • If you focus on increasing SAT scores in your freshman class, you'll pretty much assure yourself of enrolling fewer low-income students
  • But if you have higher mean freshman test scores, you'll see higher grad rates
In other words, test scores are income; income is fewer barriers to graduation.  And colleges are thus incentivized not to enroll more low-income students: It hurts important pseudo-measures of quality in the minds of the market: Mean test scores, and graduation rates.

If  you're interested on a much deeper dive on this with slightly older data, click here. Otherwise feel free to play with the visualization below.


Thursday, March 29, 2018

How have admit rates changed over time?

Parents, this one's for you.

Things are different today, or so everyone says.  If you want to see how admit rates have changed over time at any four colleges, this is your chance.  Just follow the instructions and take a look to compare how things have changed over four years.  The view starts with four similar midwestern liberal arts colleges, but you can compare any four of your choice.  (And before you ask, 2016 is the most recent data available in IPEDS).

And, a note: These changes are not all driven solely by demand.  Colleges can manipulate overall admit rates by taking a larger percentage of their class via early programs, and admit rates in those programs can be as much as 30 points higher than in regular decision.


Tuesday, March 13, 2018

Early Decision and Early Action Advantage

There is a lot of talk about admission rates, especially at the most competitive colleges and universities, and even more talk, it seems, about how much of an advantage students get by applying early, via Early Decision (ED, which is binding) or Early Action (EA, which is restrictive, but non-binding).

I license the Peterson's data set, and they break out admissions data by total, ED, and EA, and I did some calculations to create the visuals below.

Two important caveats: Some colleges clearly have people inputting the data who do not understand our terminology, who don't run data correctly, or who make a lot of typos (a -500% admission rate is probably desirable, but not possible, for instance).  Second, not every university with an EA or ED option (or any combination of them, including the different ED flavors), breaks out their data.

Start with the overall admit rate.  That's the one that gets published, and the one people think about. It's the fatter, light gray bar.  Then, the purple bar is the regular admit rate, that is, the calculated estimate of the admit rate for non-early applications (this is all applications minus all early types).  The light teal bar is the early admit rate: ED plans on the top chart, and EA plans on the bottom.  Some colleges have both, of course, but most show up only once.

You can use the filter at right to include colleges by their self-described level of admissions difficulty.

Working on another view to show the number of admits scooped up early vs. regular.  Stay tuned.  Until then, what do you notice here?  Leave a comment below.


Thursday, March 1, 2018

Tuition at State Flagships

The College Board publishes good and interesting data about college tuition, including a great table of tuition at state flagship universities. (I realized while writing this that I don't know how a university is designated a state flagship.  Maybe someone knows.)

There is some interesting stuff here, but I'll leave it for you to decide what jumps out at you: If you live in North Dakota, you might wonder why South Dakota has such low tuition for non-residents.  If you live just outside Virginia or Michigan, you might wonder why it costs so much to cross the border.

Anyway, using the tabs across the top, there are five views here:

Maps

Four maps, showing (clockwise from upper left) in-state tuition, out-of-state tuition, non-resident premium index (that is, how much extra a non-resident pays, normalized to that state's in-state tuition), and the non-resident premium in dollars.  Hover over a state for details.  You can change the year, and see the values in 2017 inflation-adjusted dollars, or nominal (non-adjusted) dollars.

States in Context

This arrays the states by tuition over time.  Use the highlight functions (go ahead, type in the box; you won't break anything) to focus on a region or a specific state. You can view resident or non-resident tuition, adjusted or non-adjusted.

Single Institution

Just what it says.  The view starts with The University of Michigan, but you can change it to any state flagship using the control at top right. Percentage increase is best viewed in 2017 adjusted dollars, of course.

Percentage Change

Shows change of in-state tuition by institution over time.  The ending value is calculated as a percentage change between the first and last years selected, so use the controls to limit the years.  Again, highlight functions put your institution in context

Non-resident Premium 

This shows how much extra non-residents pay, and trends over time.  Again, highlighter is your best friend.

Feel free to share this, of course, especially with people who are running for office in your state.

And, as always, let me know what you think.






Monday, February 26, 2018

College Board AP Data

The College Board recently released data on its AP Exams.  I've downloaded several workbooks already, and of the one I've dug into, I've only been able to get through two worksheets.  The data presentation is clunky (please, agencies, provide un-pivoted data without merged cells and totals and all that stuff, if not by itself, then as a companion), but it reveals some interesting patterns.

Well, I think so.

I've visualized it in five views: The source of the data is here, in case you want to download it yourself.

View 1, Totals (using the tabs across the top) is just totals: Use the controls to show males or females, or certain scores, or certain exams.  I think it's very compelling, especially if you look at the high scores the College Board claim about AP opening access to selective institutions.

View 2, Scores by Ethnicity and Exam, shows score distributions of the four largest ethnic groups.  Filter by a single exam if you'd like.

View 3, 100% Stacked Bars, shows the same data, presented by ethnicity.  Again, filter to a test if you'd like.

View 4, Mean Scores by Ethnicity and Exam, arrays all tests, and breaks out mean scores (yes, I know you shouldn't take averages of string variables.  So sue me).  Use the highlighter if you'd like to make any of the groups stand out visually, and filter by gender if you'd like.

View 5, Mean Scores by Gender and Exam, shows the differences between males and females. Filter to a single ethnicity if you'd like.

Tell me what you see.  Does this change your perspective on the College Board claims, or does it strengthen them?  Does it help you make up your mind?

I'd love to hear.


Wednesday, January 31, 2018

How is College Enrollment in the US Changing?

College enrollment is down.  Or maybe it's up.  Or maybe it's both.

When you read headlines, you don't get a lot of nuance. And in a country as big as ours, with such an incredible diversity of programs and widely divergent institutions, nuance is important.  So this may help do the trick.

This is enrollment data from about 6,600 post-secondary institutions in the US, and goes back as far as 1980.  It includes every institution, including those that grant degrees, and those that don't; four-year private, not-for-profits, for-profits, and publics; liberal arts colleges, research universities, and technical institutes.  All here.

It's on two dashboards.  The first shows all undergraduate and graduate enrollment at all these institutions, since 1980.  (Note: The data skips from 1980 to 1984, and I took out two years of data--1998 and 1999--because they looked a little funky.)

On the first dashboard, there are several controls to filter the data.  So for instance, if you want to look at just doctoral institutions, you can do that.  Just colleges in New England? Yes.  Only care about full-time enrollment? Just use the filter to select it.  If graduate enrollment is your interest, it's easy to get rid of the undergraduate data.  Just use the controls.  The top chart shows raw numbers, and the bottom chart shows percent change over time.  If you want a longer or shorter window, there's a control to limit the number of years.  This is especially helpful to show percent change.

Then, you can break out what ever enrollment you've selected.  Use the control titled "Color Lines By" and you can split the data shown into groups. 

Try it.  You won't break anything.  You can always reset using the little reset button at the bottom.

The second dashboard (using tabs across the top) shows similar data, but you can choose an individual college.  Once you've done so, you can limit the data shown, and you can also split it out according to your interest.

Have fun.  I've found some interesting little ditties I'll be tweeting out, and I encourage you to do the same.


Thursday, January 25, 2018

A Quick Look at the NACUBO Endowment Data

Each year NACUBO releases its study of endowment changes at about 800 colleges and universities in the US and Canada.  For this post, I'm including only those institutions in the US, and only those who reported two years of data to the survey, or about 787 institutions.

Higher Education in the US, of course, is a classic story of the haves and have nots; a few institutions near the top of the endowment food chain have amassed enormous endowments, allowing them great freedom in the programs they offer and the students they enroll. In fact, the 21 most well endowed institutions control over half, or about $280B of the $560B held overall, leaving the other 766 to divvy up the remaining $280B among them; the top 93 own 75%.

What's more interesting, I think, is the astonishing endowment growth: Stanford added $2.4B to its endowment in one year.  That amount is bigger than all but 38 of these institutions' total 2017 value.  In other words, if the gain on Stanford's endowment was an endowment, it would be the 39th largest endowment in the nation.  And in total value, it still trails Harvard by about $12B.

A couple of notes: Endowment growth is not the same as investment performance.  Some of the growth or loss can be accounted for by additions and withdrawals as well.  Second, endowments are not a big pot of money the college can spend as it wishes.  Some percentage of the income from endowments is restricted to certain programs, and often carry additional expenses the college has to come up with on its own.

Still, I think this is interesting and compelling.  Let me know what you think.




Monday, January 15, 2018

National Trends in Applicants, Admits, and Enrolls, with Draw Rates

If you read this blog regularly, you'll know I'm interested in the concept of the Draw Rate, a figure seldom used in college admissions.  Many people, when thinking about market position in higher education use selectivity or admit rate (the percentage of applicants admitted), or yield rate (the percentage of students offered admission who enroll) by themselves.

But in the market of higher education, these two variables often fight against each other. (BTW, if you object to the use of the word "market" in higher education because you think it debases our profession, see what Zemsky, Wegner, and Massy have to say about that here.)

Colleges, driven by market expectations, have for a long time tried to increase applications, believing that what the market wants is greater selectivity in the institution they choose, based on the Groucho Marx effect. Except that in order to enroll the class you want, you have to take more students when apps go up (at least in the case of the bottom 90% of colleges).  That's because your incremental applications almost certainly have a lower propensity to enroll.

So, Draw Rate (yield rate/admit rate) helps account for that.  Higher Draw Rates are generally a sign of higher market position.  Think about it mathematically: A very high numerator (high yield) coupled with a very low denominator (low admit rate) is the thing many colleges pursue.  If you pursue greater selectivity and don't account for the lower yield, you won't be in enrollment management too long.

The problem, of course, is that, in general, people who were not born 18 years ago don't apply to college.  And the number of people who will turn 18 in any given year continues to drop going forward.  So no matter how many applications each student makes, they can only go to one college next fall.

Over the past several years, the "Winner Take All" mentality has driven demand at the most selective institutions.  The need to keep up trickles down to each tier below, and the annual "We received a record number of applications for this freshman class" shtick gets old fast, even if colleges have not gotten that message yet.

The take away: Colleges have been spinning their wheels, working harder and harder to generate more applications just to stay even.  The national psychosis weighs heavily on the minds of parents and students, and they respond by hedging their bets, applying to--guess what--more colleges.  And the spiral spirals out of control.

Here are five views (using tabs across the top) to show the data.

Dashboard 1 is a high level overview of applications, admits, and enrolls at four-year public, and four-year, not-for-profit institutions (open admission institutions do not report application activity to IPEDS).  You can use the control at top to show all institutions, or just public or private.  Top view is raw numbers; bottom is percent change.

Dashboard 3, the next tab, shows the same data on bar charts, with the draw rate as a brown line hovering over the bars.  Note how it's dropped over time: This is the effect of soft applications.  You can look at any region, or any single institution if you want, but the really interesting filter is at top right: Compare colleges by their 2016 selectivity.  You see that the only institutions who have collectively increased their draw rates are exactly the ones who had the strongest market position already: The most selective colleges.  Step down from Most to Highly to Very, etc, and watch the trend on the brown line.

Next comes Dashboard 2, showing Applications per Seat in the Freshman Class, and draw rate by region.  This might explain why we in the Midwest are fascinated with the obsession with college admissions by East and West Coast media.  Y'all are welcome to come to the Midwest and chill, if you'd like.  You can use the filter to select groups of colleges by Carnegie type.

Dashboard 4 shows four key metrics to reinforce the relationship between and among them.  Again, select by 2016 Selectivity to see how they make a difference.

Finally, Dashboard 5 allows you to compare individual institutions.  I've put Harvard, Stanford, and MIT on to start, but you can choose any colleges you wish.  (I recommend no more than three or four at a time.)  To remove a college, hover over its name in the filter and X it out.  To add, type any part of the name and hit "Enter" on your keyboard.  You'll be presented with all possible matches, and just choose the ones you want.  I recommend choosing similar institutions for scaling/charting purposes.

I hope this is interesting to you; let me know what you see, and if you spot any problems.




Wednesday, January 3, 2018

Freshman Migration, 2010-2016

This is perhaps the most popular, as well as my personal favorite, post, and I'm sad that I can only do it once every two years (as the IPEDS reporting cycle only requires this data be reported bi-annually.)
This shows patterns of freshman migration within and outside of state boundaries. It's valuable to people because you can see the composition of freshman classes at colleges: Where do the students come from? You can also see patterns of state exports: Which states keep students at home, and which send them out-of-state (of course, the size and educational offerings of the various states means it's often unfair to compare, but it's still interesting.)

For this, I've limited the universe to four-year, public and private, not-for-profit institutions. Community colleges and for-profit colleges tend to have very local enrollment patterns, and high numbers of part-time students. I've also taken out institutions whose primary focus is religious training, as well as those from a few obscure Carnegie categories.

The freshmen in this analysis are only those who graduated within twelve months of enrollment in college. A word of caution: If you are afraid to click buttons and interact, stop now. This won't be of any help to you. You can't break these, and you can always reset using the controls at lower right. So click around and explore the data.

Finally, this shows the data I downloaded. Some of it is pretty clearly wrong, but that's not my problem. Contact the IR office at the offending institution and ask them what they were thinking.

So, first up: If you want to compare any four colleges on the geographic composition of their freshman classes, start here. I've added four colleges that start with "D" but you can use the controls to look at any four you want. Note: Students labeled as "in-region" are from the region, but not the state. Therefore someone "in-region" in a New Hampshire college would be from one of the five other New England states. Got it? Good. Play away on this one:



Next up: Looking at the bar charts: It's a little more complex, but you can do it.  If you want to see which colleges enroll the most (top chart) or highest percentage (bottom chart) of students from in-state, in-region, or out-of region, this is your visualization. Choose a year (it defaults to 2016), and if you wish, limit it to colleges in a region (The Southeast, for instance).  You can limit to public or private as well.  Then choose which group of students you want to explore: In-state, in-region, or out-of-region.  Again, comparing Texas to Rhode Island should only be done for the "interestingness factor," not to draw conclusions.



Here is the same data, represented on a scatter plot, in case you want to step back, and see the data all at once.  The two scales are the number of freshmen, and the percent from the region selected.



Which states export the most students, and when they export them, where do those students end up?  If you've wondered that--or if you're from Illinois or New Jersey and lament our students' mobility--this is the visualization for you.

Choose a year, and see (on the top bars, in purpley-mauve) which states exported the most students.  Then, click on a bar representing a state to see where students from that state enrolled, in the bottom chart.  If you want the college destinations to be limited to public or private, or a certain region, you can use those controls to do so.



And finally, if you're interested in which states keep students at home, you can see that, too, on this visualization. The top view looks at colleges in a state, and where their students come from; the bottom looks at students from that state, and whether they go out-of-state or stay in-state.  Again, choose a year or institutional type, if you want to look at colleges or students going to those types of colleges.



I hope you have enjoyed looking at this data as much as I have enjoyed playing with it. If you spot any errors that I've made (Tableau still has no spell check....) let me know, and I'll get to fixing them right away. Otherwise, leave a comment below with questions or observations.

Wednesday, December 13, 2017

How Many Colleges are There in America?

Seems like an easy question: There are 7,284 post-secondary options in the US.

But everyone has a different definition of what they want when they ask for a count of colleges.  This should give you some clearer sense of the right answer for you.

At top left is "The Answer," and that will not change as you navigate through this.  But you can use the controls here to change the number of colleges and universities you're looking at, and to change how they're broken out.

Those controls change the number (in orange, at top) and the splits.

For instance, at the far right, on the control labeled "Region, choose "Great Lakes," and you'll see that there are 1,079.  On the gray box at top right, choose "State" and you'll see 354 in Ohio.  Under "Control of Institution" choose "Public" and you'll get 266.  And so on.  Now break out by "Campus Location" and see most are located in cities.

The reset button is at lower right.

I hope this is helpful to you as you wonder about the shape and size of American higher education.


Monday, December 11, 2017

What's All The Fuss About, Redux

My tireless crusade continues.

Everywhere you look, it seems most of the discussion and ink spent on higher education focuses on the most selective institutions in America.  In addition, if you listen to parents and students and counselors talk, you'll learn that there is a perception that college is increasingly hard to get into.

So, I broke the whole world of 1.403 four-year private, not-for-profit and public colleges and universities into bands, based on the absurd input measure of their freshman selectivity.  On the visualization below, they range from red (less than 15% admitted) to purple (over 60%) admitted.

Each institution falls into one of these boxes.

The four charts, clockwise from top left: The number of colleges in those categories, the number of freshmen they enroll, the total number of freshmen with a Pell grant, and the total undergraduate enrollment.

If you think you see a lot of purple, you do.  And this is before anyone enforces any sort of standard definition of what an "applicant" is.  Sometimes, it's just a person who accidentally clicks on an email link.

Of course, sometimes the scarcity of a good is exactly why people freak out about it. And of course, this doesn't even consider open admissions colleges (nine percent of all college enrollment in the US is in California's Community College System). So, this won't change the world, but I feel better for sharing.  Now you can't say you weren't told.


Friday, December 1, 2017

2016 IPEDS Admissions Data

Fresh from IPEDS, just months after the wrap up of the 2017 admissions cycle, comes the 2016 admissions data.

I've done something a little different this year to focus your attention, using five views of data, navigable via the tabs across the top of the visualization:


Admissions data (first tab) is pretty clear.  Colleges display admit rates (overall, in red) and then admit rates by gender (men are in blue; women are in orange).  If the blue bar extends beyond the orange, you can see that the admit rate for men is higher, and vice versa.

On the right are standardized test scores, showing calculated means.  In other words, since no one publishes averages and everyone wants them, I took the mid-point of the 25th and 75th percentiles to approximate the 50th percentile.  Note that IPEDS does not allow colleges that are test-optional to report test score information.  Also note that I've taken out a lot of colleges with extremely limited or suspect data.

As always, you can play with the filters (if there are any) to limit the colleges displayed, and you can sort columns by hovering until you see this little icon, and then click on it.

You can reset the view by clicking this little icon at lower right.

The four other views show a limited scope of colleges: Selective, wealthy, mostly men, and Land Grant institutions, and plotted them using some variables that should both answer and generate questions.

You be the judge. 



Tuesday, August 15, 2017

Chasing the Endowment Unicorn

Higher education is struggling these days, and there are a lot of solutions from a lot of pundits, all of which tend to be macro in nature: Delivery, cost structures, optimization, curricular adaptations, and many other ideas abound.

On the micro level., however, the vast majority of the 1,700 or so private, four-year colleges and universities will point to "increasing our endowment" as one of the most crucial solutions to our internal institutional challenges.

This is, in all probability, because the wealthiest institutions in the nation (in terms of endowment resources) are also the best known, and much of the brand of any institution is driven by wealth and reputation and prestige.  And even in this decade and these trying times, some of these institutions have parlayed considerable investment income into one-year operating surpluses of over a billion dollars. No, that's not a typo; it's a problem every university president would love to have. (Reminder to self: Update this chart.)

I once had a finance professor suggest that every institution should multiply the amount of money spent on Advancement each year by 20, then consider these options:

Let's say your Advancement Office budget is $8 million per year.  It would take an endowment increase of about $160 million to throw off that $8 million in cash each year forever (at 5%). Thus, shutting down the Advancement function completely would be the equivalent of raising $160 million in unrestricted endowment overnight. Unrestricted dollars are the hardest to raise, of course, because people don't tend to say, "Here's five million dollars; do with it whatever you want."

(It's also a good time to remind people that much endowment money is restricted; the $20 million gift from a big donor doesn't usually provide general operating relief but instead is used to fund some center or institute or faculty chair the donor thought was a good idea.  So in some sense, total value of the endowment can be occasionally misleading. It's still generally better to be bigger, though.)

Due to head starts and compounding, the wealthiest institutions are so far ahead of the rest of us that even trying to catch up seems futile.  Of course, that stops no one from relying on the old "tried and true."  In reality, our only hope of catching up with them would be a catastrophic market crash with no rebound; even then, we'd all be poor.  No solace there.

Take a look at the interactive visualization below.  Each bubble is an institution.  Hover over a bubble for details.
  • The SIZE of the bubble indicates endowment value at the end of FY 15 (probably June 30, 2015)
  • The COLOR of the bubble indicates tuition dependency (in IPEDS, "Percent of core revenues from tuition and fees.) Orange is low; blue is high.
  • The relative position on the y-axis (up and down) indicates one-year endowment value change (note: This is just subtraction, so it is not endowment performance).
  • The relative position on the x-axis (left and right) shows the one-year percentage change.  I cut it at 50% each way for clarity as there were a few extreme outliers.
If you'd like, you can use the filters at the top right to limit the types of institutions shown, or the range of endowment values.  Use the highlighter at the top left to highlight a specific institution.  Just start typing any part of the name to do so.

How do you feel now?


Saturday, July 8, 2017

Changes in College Attendance by State and Ethnicity, 2005-2015

Note: If you haven't read my post about the 2016 election results and educational attainment, it might be of interest to read that first.  Or later.  Or not at all. Your choice.

This started simply enough: A couple of tables from the Digest of Education Statistics, (tables 302.65 and 302.70) showing the percentage of adults aged 18-24 who were attending a degree-granting college by state and ethnicity in 2005 and 2015.  If you've read this blog enough, you know I have a love/hate relationship with the digest: Great data, but horrible formatting.  The tables are made to be printed on a single 8" x 11" sheet and handed out.  The crucial distinction between data and insight is lost.

Regardless, I reformatted the sheets into something workable for Tableau, and started to look at them. I wasn't having much luck: Some of the states didn't have data on African-American students, for instance, in 2005.  The variable for "Asian/Pacific Islander" was relatively new then, and only a few states had that data available.  Beyond that, I was looking to add some color-coding into the visualization to help make a point, and it wasn't going well.

But I've been fascinated since the election by some of the tweets and writing of Chris Arnade and Sarah Kendzior, who are thinking about what the election results mean in "flyover land."  And my blog post about the election results and attainment has stuck with me, mostly because of the reaction people had to it.

So I colored the states by the 2016 election results, and it got more interesting, as you can perhaps see below.

It's easy for us to look at things like this and chalk it up to "uneducated people voted for Trump." While that may technically be true, leaving it at that makes it too convenient for us in higher education to forget that educational attainment is only partially something you earn; it's also something you're born into.  Some of the ten charts on this post might make that clearer.

This can also, of course, be a post about urban and rural, divides. The division in our country might be as much about opportunity as it is about attainment.  If history tells us anything, it's that people start to rebel when they feel they don't have a chance via any other path.

So as we look at the current reality, the question, as always, remains: What are we doing to change the future?



Thursday, June 29, 2017

What Happens if Federal Money Goes Away?

Strategic planning at universities is always an important process, but it's even more crucial to do correctly these days.  And lots of institutions might be missing a really critical element in scanning the external environment: The extent to which federal financial aid programs contribute to the essential revenue streams that run the enterprise.

This is a fairly simple, if crowded, visualization, showing about 900 private colleges and universities who have good data in IPEDS.  Each dot is a bubble, colored by region, representing a single institution.  Its position along the horizontal axis shows student loans as percentage of core revenues, from left (low) to right (high).  I've included subsidized undergraduate direct loans, unsubsidized undergraduate direct loans, Parent PLUS loans, graduate subsidized, and Graduate PLUS loans in the calculation.  I did not include private loans.

Some of these numbers may seem high, but understand what this says and what it doesn't say: Loans go to pay other things (computers, gasoline, rent, books, food, etc.) so the colleges don't actually see all this money.  But presumably, the funding does make attendance and the paying of tuition possible.

And the IPEDS definition of Core Revenues can be confusing, too, as there are many revenue sources you might not consider.  This is what IPEDS puts into the category of Core Revenues:


  • Tuition and fees revenues (F2D01) 
  • Federal appropriations (F2D02) 
  • State appropriations (F2D03) 
  • Local appropriations (F2D04) 
  • Federal grants and contracts (F2D05) 
  • State grants and contracts (F2D06) 
  • Local grants and contracts (F2D07) 
  • Private gifts, grants, and contracts (F2D08) 
  • Contributions from affiliated entities (F2D09) 
  • Investment return (F2D10) 
  • Sales and services of educational activities (F2D11) 
  • Other revenues (F2D15) 

And if you have investment losses, your core revenues drop.  In other words, it can be misleading. And even if it doesn't, most places don't spend all of their investment returns, so while it shows up as a revenue, it is usually never touched.

Got it?

Second, on the y-axis, is Pell Grant revenue as a function of your Core Revenues.  Same idea as above, but using Pell as the numerator over Core Revenues.

Add these two together, and you'll see what happens to your revenue stream if federal aid goes away.

The bubbles are sized by tuition dependence; the calculation is not standardized, so for the sake of simplicity, I looked just at tuition revenue as a percentage of tuition plus investment income.

If you want to show a single or a group of institutions in context, use the filter.  Just type part of the name and select it.  If you want to look at fewer institutions, choose a region, a state, or reduce the range of core revenues (for instance, type $100,000,000 in the left hand box of the filter, or use the slider, to eliminate very small institutions.)

As always, hover over a bubble for details.

You'll notice some interesting things, I hope.  Mostly, I hope this doesn't frighten you.  Depending on where you work, it can be a bit daunting.


Friday, June 16, 2017

The Discount Dilemma

"You should write something about discount rates."

I hear that a lot these days.  Even though NACUBO does its annual discount study, people still want and crave more.  There is no topic, it seems, as much on the minds of people in universities as discount rate.

But despite my desire to make you, the loyal readers of this blog, happy, there are a lot of reasons I haven't written about discount rates:

  • First, data are old.  It's a long story, but financial reporting (where you learn about financial aid) is reported about a year after the freshman class enrolls.  So the ability to calculate discount is always behind the most current admissions data.  The viz below is for 2014 freshmen for instance, and it's the most recent publicly available.  It's hard to describe to people how much things have changed between 2014 and 2017. (And even harder to figure out why 2016 admissions data are not out yet.)
  • Second, discount is not as important as accountants think it is.  "WHAT?!?" say the accountants! I politely suggest that what you really care about if you're running an institution is net revenue per student, and total net revenue, the cash you use to run the university.  
  • Put it this way: If your tuition is $50,000 and you have a 40% discount rate, you net $30,000 cash per student.  But you're generating less net revenue than your competitor who charges $55,000 in tuition with a 43% discount rate, who nets $31,350 per student.  Still, people insist on comparing disparate institutions on this single accounting measure. Lower discount is not always better.
  • In that same vein, average and total revenue are both important.  If you took only full-pay students in the scenario above, you would average $50,000 with a zero percent discount.  But your numbers of enrolling students would go way down, as would your total; you wouldn't have enough to cover your overhead.  Inversely, you can generate more total revenue by discounting more and enrolling more,  but your costs go up faster than revenue, which is of course not good.
  • Third, most people and stories focus on freshman discount.  If you're a small, tuition-driven, liberal-arts college, that might be meaningful, as freshman may make up 25% or 35% of your revenue. But it's less important at big, complex universities, where we have freshmen, transfer, law, medical, graduate, and other types of enrollment too.  Overall discount is far more important, but usually less discussed.  At some institutions, where educating undergraduates can almost be said to be a side business, the revenue from student tuition is tiny, dwarfed by things like research dollars and endowment return.  Large discounts on tiny fractions don't add up to much.
  • Fourth, net revenue is not the cost to the student.  Pell and state grants come to the university as cash, and look the same as money from a student's pocket, even though they are very different to the student.  You can't tell how much a college costs a student by looking at this data.
  • Fifth, not all discount rates are the same, even at similar institutions with similar tuition rates. You can get to a 50% discount rate by having the class half full-pay and half-full need that you meet; you can also get there by having everyone at 50%.  These are impossible examples, of course, but you get the picture.
  • Finally, discount rate--an accounting measure--is something we used to look at only after the fact, where three decimal points are very satisfying.  It's essentially impossible to manage to that rate in a fast-changing, dynamic environment unless you constrain other outcomes.  As I've said before, when we send out aid awards, we're not planting saplings; we're casting seeds.  You can predict with some precision what percentage of which seeds will take root and grow, but you can't control the wind, the rain, or the temperature, which are all critical to success.
As you look at this, there are a couple of things to consider in addition to the usual caveat about the accuracy of IPEDS.

I calculated discount by taking the Financial Aid Cohort of freshmen (first-time, full-time, degree-seeking) students and multiplying that number by the tuition, then using the total institutional aid.  At most places, institutional aid is not funded, it's just a contra revenue.  But at others, income from restricted endowment funds may actually fund the aid.  Here's a long boring argument about whether this matters or not. For this purpose, I'm taking all aid as unfunded discount.  Thus, discount = Institutional Aid/Gross Tuition Revenue.

Some numbers seem a little crazy.  While a lot of institutions have freshman cohort numbers lower than total freshman enrollment (which makes sense) some have freshman total numbers lower than the aid cohort, which suggests someone in the IR office is counting wrong.  Not my problem.

This does not include all 2,200 four-year, degree-granting private colleges and universities.

Discount only makes sense for private universities, for one thing, which takes us down to about 1,600.  Several hundred of them don't accept freshmen, others have incomplete data, some are rabbinical institutes or schools of theology or massage therapy that aren't of great interest to many. Others have tiny freshman classes.  Thus, this view starts with 675 institutions, and shows five things. The four columns, left-to-right show Calculated freshman discount, Average Net Revenue per Freshman, Endowment Per Freshman, and Percentage of Freshmen with Pell.  The color shows mean SAT CR+M scores, approximated using the 25th and 75th percentiles; gold is low, and purple is high.

Use the filters on the right to limit the view.  For instance, you might only want to look at colleges in your state, or with similar sized freshman classes, or test score averages.  Or perhaps you only care about Doctoral institutions.  Filter to your heart's content.  If you want to sort the columns, just hover over the bottom of the column until you see the small icon appear, like this:

Click through the cycles to sort descending, ascending, or alphabetically.  Click on undo, redo, and reset to, well, you get the picture.


Let me know what you think and what you see here.  That is, if you've gotten this far and I haven't whomped the enthusiasm right out of you.




Tuesday, April 4, 2017

Undergraduate enrollments by ethnicity, 2015

I was doing some research for our own internal discussions, and decided to take it a few more steps to look at enrollment of undergraduate students by ethnicity at about 2,000 four-year, public and private institutions in the US.  (And when you look at the data and wonder why, rest assured, I checked: Miami Dade does offer Bachelor's degrees via online programs.)

It's here, and the first two views are pretty easy to navigate:  Each chart shows a separate ethnicity and lists each institution in descending order.  The first view is by counts, and the second by percentages.  Thus you can see the institution that enrolls either the most Hispanic students, for instance, or the institution with the greatest percentage of Hispanic students, depending on your preference.

If you'd like to focus on a single state, just public or private, or colleges of a certain size range, use the filters at the top.  You can always reset the views using the control at the bottom.

The third view allows more customization.  Each point represents an institution, arrayed on the x- and y-axis.  But you can control what values the axes show: For instance, percentage White on the x-axis, and percentage Asian/Pacific Islander on the y-axis.  The points are colored by control: Orange is for private institutions, purple for public. Again, you can limit by undergraduate enrollment or by state, if you'd like.  But this view has the advantage of choosing a highlight institution: Use the highlight box to put a university of interest in context.  Type part of the name, and select it, and it will show up all by itself.

I hope this is helpful for use with students who are interested in thinking about and comparing colleges and universities by enrollment profile.  And if you're interested in seeing how an ecologist might look at enrollment diversity, check out this piece I wrote for Academic Impressions last fall.


Friday, March 10, 2017

Is This Why Democrats Support Education Funding?

This post started off simply enough: I found some cool data on changes in educational attainment over time. I was going to take a look at how far we'd come as a nation in the last 40 years (even though I had already published this), and show where the biggest gains were.

It wasn't very compelling, at least at first.

Then, I decided to get ambitious (my wife was in her evening class, so I had the night free), and wondered if there were any interesting connections between changes in educational attainment and voting patterns in 2016.  I found a data set with election results by county, (a handful of counties are missing) and merged it in. And thus, this.

Before I start, there are a few points to make about the data.  First, the definitions changed slightly over time.  For instance, in 1970, the field is labeled "College degree," while in 2010, it's labeled "Four or more years of college."  Not the same thing, but we'll have to go with it for now.  Also, 2010 is not really 2010; it's the data from the five year American Community Survey of the Census Bureau, but there's no reason to believe it's not as accurate as the census itself.  In fact, the ACS is used to test the accuracy of the census.

So, onward.

Using the tabs across the top of the visualization:

First, a scattergram, plotting attainment in 1970 and 2010.  The regression line suggests that attainment has essentially doubled in 40 years; those bubbles (counties) above it have done better; those below worse.  Bubbles are sized by votes in the county in the 2016 presidential election.  If you want to look at just one candidate, use the highlighter function.

Note that the counties that went for Clinton tend to be larger (more urban), with higher levels of attainment (moving toward the top right) and more above the line.  Counties that went for Trump tend to be the opposite.  And of course, there are many exceptions (Johnson County, Kansas; Apache County, Arizona), and clearly the binary blue/red can be misleading; some victories are by a point, some by 15 points or more.  Finally, it's almost certain that much of the change in attainment is due to people moving in and out; not everyone lives where they were born. But it's still interesting.

Second, the bar chart (I know from experience that many people won't click on the second tab. Please. You will be glad you did).  The x-axis is broken into college degree attainment. For instance, the long bars in the center show counties where 30--34.99% of adults have a college degree.  You can see how many votes these counties cast for Trump, and how many for Clinton.  I double checked this; it is perhaps the best story I've ever told with one chart.  And although the left end changes as you select single states (using the filter at the top), the right end is fairly stable.

Finally, the last chart just shows three variables: 1970 and 2010 college-degree attainment, and the change over time.  See the box to the right of the chart if you want to sort the data.

Admittedly, this election was a strange one, so perhaps there are no lessons to be learned.  But over the past few decades, Republicans have been fairly staunch opponents of increased educational funding, and you have to wonder if this doesn't explain why; people who lived in areas with higher levels of education voted for Democrats in the last election.

Fifty years ago, the Republicans were the party of the college-educated, white collar classes; the Democrats the blue collar, working-class, high school educated citizens.  That's all changed, if 2016 is any indication.

Agree? Disagree? Let me know in the comments below.


Friday, March 3, 2017

2016 Freshman Admissions Data

Note: I've just discovered that although this data set is labeled 2016, it is for the 2015-2016 data year; thus, this is Fall, 2015 admissions data, not Fall 2016 as I had thought.)

This always proves to be a popular post: The 2016 Admissions Data summary.

Here you'll find ten views, showing test scores, admit rates, need data, and international student information (which should only be used as a guide, as you'll see.)

Use the gray boxes and/or arrows across the top to navigate this information, and the filters to limit the views.

Note: This data is "as reported" to Peterson's and is presented as is.  If it's wrong or your college is missing, it's almost certainly a reporting error; most institutions left at least some fields blank.

It comes from the Peterson's Undergraduate database and the Peterson's Undergraduate Financial Aid database, both copyright 2016 by Peterson's-Nelnet.  The data here are used with permission of Peterson's.