Insights into Online Education and OPMs from the College Scorecard

, /4 Comments/by

Most public discussion around program-level College Scorecard data for student debt levels has focused on transparency in terms of helping students decide which schools to consider, but I think a more important potential benefit is for policy makers, institutions, EdTech vendors, and even investors. Better analyzed data should inform people and organizations making decisions on creating, funding, and supporting academic programs. A case in point is the insight that you can get by looking at a school that offers online programs, whether using an Online Program Management (OPM) partner or not.

Simmons University is a small private school in Boston, with women-focused undergraduate programs (fewer than 1,800 students) and co-educational graduate programs (similar numbers). Despite its small size, the school has two of 2U’s largest programs (Nursing and Social Work). Simmons has other programs signed up with 2U that were not active as of the beginning of July 2015.

Interestingly, Simmons also has an online Library Science program that does not use 2U or any other full-service OPM partner.

I combined the College Scorecard data with IPEDS and manual augmentations from public 2U and Simmons University websites to get a deeper look. Some notes first to better understand the data:

  • This Department of Education data set is preliminary and has not been reviewed by colleges and universities. The ED plans to release a final version in Fall 2019.
  • The data has been combined into two-year cohort groupings with the data below from July 2015 through June 2017.
  • I have listed below whether Simmons offered fully-online options for each program as of the 2015-17 cohort, but there are also on-campus versions of the same programs, and student counts and debt levels are aggregated.
  • The number of students is taken from the two-year cohort award grouping and not from the IPEDS student counts.
  • Update: The student count is the number of borrowers in the cohort, not the total number of students. Students who were enrolled during that cohort period but took out no Title IV-based student loans would not be included. For example, Simmons’ master’s of social work has a Count (shown below) of 440 but an IPEDS-derived cohort enrollment of 534, and master’s of nursing has a Count of 532 but an IPEDS-derived cohort enrollment of 684. The Median and Mean data goes across the borrower count.
  • Some programs do not show student debt levels due to this information being suppressed by the College Scorecard for privacy reasons (e.g. too few students would allow indirect determination of student identities).

One of the reasons that I chose Simmons University and 2U 1Disclosure: 2U is a past client of MindWires and is a sponsoring participant of Michael’s Empirical Educator Project. as an example is that both organizations are remarkably transparent with their data. 2U is not the only publicly-held OPM company, but they are the main one where OPM is their primary business (unlike Pearson or Wiley), but even so, they provide more information than any other company. To keep this as an exploration of public data, I have not contacted either Simmons or 2U for commentary.

Simmons University Grad Programs by # Students and Median Student Debt Levels and Online Offerings

Some notes from my initial exploration of this data:

  • Tuition for the graduate programs ranges from $980 to $1,450 per credit hour with ~40 – 60 credit hours required for a degree.  This leads to total tuition in the $50k – $70k range. But look at the median student debt levels, most of which are in the same rough range as total program tuition. Does nobody use their own money for graduate school? (Update: see note above about the Count data being for borrowers only. In Simmons’ case, roughly four in five students were borrowers.) Education seems to be the only academic discipline with student debt significantly below program tuition.
  • This data makes it more clear why a school might offer online programs. The three largest graduate programs, by far, have online offerings, and I would assume that the vast majority of students are online in these programs. For Simmons going online means the opportunity to move from dozens of students to hundreds of students per program. I realize this should not be the only, or even primary, reason for a school to offer graduate programs, but the data make this factor quite evident.
  • The two largest programs offered online are in partnership with 2U, and according to investor analysis, both Simmons and 2U make $10 – $20 million in tuition revenue per year.
  • The third largest program offered online, library science, is the only one at Simmons not using 2U or any OPM, at least that I can figure out. This makes an interesting case that I’d like to better understand. Why does this program appear to be managed internally and how did it get up to hundreds of students per cohort?
  • At the same time, the two large 2U programs have the highest student debt levels. Is this a feature of the disciplines involved (social work and nursing), the usage of an OPM, both, or some other reason? The debt levels are roughly 15 – 16% higher than the third placed program (in terms of median debt levels).

It  might be interesting to compare data across multiple OPMs or across different disciplines to get further insights. For now, this case study seems to provide a lot of insight into online education and OPMs and general student debt levels for graduate programs.

Update 6/10: See in-line notes clarifying how the College Scorecard defines Count. Also added emphasis that online and f2f programs are aggregated.

Visualizations of Program-Level College Scorecard Data on Student Debt

, /2 Comments/by

The US Department of Education released initial program-level data for student debt as part of the College Scorecard. The goal is to increase transparency around specific program outcomes and not just aggregated by institution. From the press release:

Finally, building on President Trump’s March 2019 Executive Order on Improving Free Inquiry, Transparency, and Accountability at Colleges and Universities, the College Scorecard released new preliminary loan debt data by field of study to the College Scorecard data webpage. For years, the College Scorecard provided undergraduate loan debt information at only the institution-level even though the amount borrowed to attend school can vary substantially depending on which program the borrower is enrolled. Institutions are being asked to update their historical enrollment data from which these loan debt metrics are derived. These adjustments will be reflected in a subsequent release of loan debt data metrics by field of study this fall. For prospective students, families, researchers and others interested in examining loan debt data by field of study, please note that the data released today are considered preliminary and future improved versions will be calculated after institutions have time to adjust their historical enrollment data.

There have been several analysis posts already about this new program-level student debt data, and I started using Tableau to combine the student debt with IPEDS data to get a better sense of what new information we have, and I thought it might be useful to share the data visualizations. Below I block quote findings from two articles and place a data visualization below each section. I have color-coded the type of institution in the following cases – blue for public, green for private, and red for for-profit.

The first analysis was from Robert Kelchen, using combined 2015-16 and 2016-17 dataset (the most recent year available).

Median debt data are only available for 42,430 programs (21.8% of the sample), as small programs do not have data shown due to privacy concerns. But based on IPEDS completions, about 70% of students are enrolled in programs where debt data are available. [snip]

172 programs had over $200,000 in median debt, and it looks like the top 116 programs are all in health sciences. The data are preliminary, but Roseman University of Health Sciences’s dentistry program has the top listed debt burden at a cool $410,213.

Median Debt Rank All Programs

Andrew Kreighbaum writing at Inside Higher Ed focused on the large student debt held by students coming out of master’s programs.

The Scorecard information is preliminary and hasn’t been checked by institutions themselves. But it to a large extent confirms what researchers already knew about student debt. Graduate programs, it shows, have the highest volumes of student loan debt. It also highlights a number of outlier programs with especially high debt numbers.

Median Debt Rank All Programs

The University of Southern California, for example, features the only master’s program in social work where average student borrowing topped $109,000. Social workers tend not to earn high salaries and are therefore likely to spend much more of their income on student loan payments.

Borrowers at more than 200 other programs listed in the Scorecard data borrowed on average less than half that amount. Only three other master’s in social work programs — at DePaul University, Howard University and Capella University — had average student borrowing of more than $90,000, according to the most recent data.

Median Debt Social Work

Among law schools with the highest average student borrowing, one of the top three, Whittier College, closed recently in part because of poor outcomes. The accreditor for the fourth, Florida Coastal School of Law, last year found the for-profit out compliance with its standards. Florida Coastal subsequently posted improved bar-passage rates and is seeking to convert to a nonprofit.

Median Debt Law

I also created a combined map view of MBA and related master’s of business degrees to get a better sense of geographic overlap.

MBA Related Mean Debt Level

This data are not perfect, but it is good to have increased visibility into actual programs of study as well as more focus on student debt levels.

Student Panels: Non-traditional students and consistency in course navigation

, /5 Comments/by

Two weeks ago I had the chance to deliver the keynote at an Athabasca University’s Community-Wide Conversation on Future Virtual Learning Environment. Athabasca is a Canadian university in Alberta with a long history of providing distance education to nontraditional student populations. “Virtual learning environment” in this case refers to an ecosystem of learning applications, and not a sole focus on the LMS.

After my keynote I had the chance to watch a student panel moderated by Provost and VP Academic Matt Prineas. When the students were asked about their biggest frustrations, or opportunities for improvement, 2One was a student council president who said his few is quite consistent with other students the panel reached a quick consensus that I shared online.

What they described was that as adults with jobs, families, and lives outside of school, it is frustrating for them to face barriers, or friction, that get in the way of their coursework. Several of my professors put their assignments under this left column menu, but for another course I have to remember that the assignments are named differently and located under the top row pull-down menu taking multiple clicks to access. Why do I have to remember different paths to find the basics that I need to complete this course? (paraphrasing here).

This is a common theme I have heard in several student panels at conferences as well as in many student interviews I’ve conducted for e-Literate TV or for consulting work. I also think it is an under explored issue given the importance of institutions moving beyond residential on-campus programs for 18 – 24 year olds. 3Yes, I know that the ‘traditional / nontraditional’ language is flawed. And this feedback also applies to hybrid courses as well as fully-online ones. Providing access to educational programs to working adults, students with families, first-generation students, and others means more than just enrollment and course registration. We need to understand the context of how students are working to get their degrees or certificates, and we should minimize barriers for them.

None of this is to argue for a rote, surface-level interpretation of student comments on a panel. We need further study and understanding of how widespread is this desire for navigational consistency, how best to improve our learning environments, and even how to meet needs without falling prey to consumer-driven reactions that are not beneficial to students. There is complexity here to understand.

Tanya Joosten, Co-Director and Principal Investigator at National Research Center for Distance Education and Technological Advancements (DETA), chimed in that she has also seen this feedback and is working on further research on the subject.

I am looking forward to reading the DETA research when it is available.

Mike Caulfield asked a relevant question about the deeper meaning of student feedback on this subject.

This is a good point that deserves further analysis. There is also a risk, however, of using this concern as an excuse to not fully listening to students, and I do not think that consistent navigation must imply cookie cutter course designs. While consistent course designs can be influenced by learning platform usage, the bigger issue there is departmental and institutional policy along with professional development for faculty.

I do not hear students asking for cookie-cutter courses, and I have had several interviews with students who also appreciate different course design and learning experiences. But there are a set of basics – syllabus, grade books, assignments, calendar, etc – that if the course uses them, then they shouldn’t be hard to find. And hard to find has implications across a program, not just a course.

There is an added benefit to consistent navigation in addressing accessibility concerns and moving towards Universal Design. The University of Arkansas at Little Rock summarized in a valuable post “Ten Steps Toward Universal Design of Online Courses”, with the second item as “Provide simple, consistent navigation.”

The broader message goes beyond course navigation in an LMS, of course, and it gets into the need to better understand the learning experiences of nontraditional students and the need to listen to students more often. Kudos to Athabasca University for a valuable event and specifically for the student panel.

While I will not be able to attend, I am looking forward to hearing from the student panel at the upcoming Online Teaching Conference to be moderated by Pat James and Kevin Kelly. Consider this post as my attempt to influence the questions.

USciences eLearning Conference: Keynote on future of learning platforms

/1 Comment/by

At the recent University of the Sciences eLearning 3.0 conference in Philadelphia, held in conjunction with the Northeast eLearning Consortium, I had the opportunity to present a keynote on the future of learning platforms. Rod Murray, publisher of Rod’s Pulse Podcast, hosted the event, and he just put up a video of the presentation as a video podcast. The webpage shares a description of the event along with a YouTube video, and the podcast page shares a straight audio version. Rod’s podcast is a valuable resource for the community – make sure to check out other episodes.

The event had some very interesting sessions and even finished with a craft brew social, including a tasting of two beers brewed as part of the USciences Brewing Science program.

As for the keynote, I tied together many of the themes I have shared by blog and in our LMS Market Analysis service into a (hopefully cohesive) argument that if you want to understand the learning platform of the future, you need to first understand the history of the LMS. Unlike most EdTech systems, the LMS is characterized by mainstream adoption as an enterprise system dominated by a concentrated market around the Big Four – Canvas, Blackboard, D2L, and Moodle. This situation has surfaced the primary drivers of the overall market, and by understanding these drivers we can better estimate the future path. This is a different situation than for VC-driven EdTech markets characterized by sporadic adoption mostly by non-mainstream schools, programs, and faculty.

There is a valid argument that this future market that is tied to the past can hold back innovation, but there is a counter-argument that the market actually reflects that institutions and faculty  – at least the mainstream – needs and asks for. The keynote focuses more on understanding the drivers of how we got here and what the market is capable of delivering, regardless of merit.

I couldn’t avoid doing a little soapbox discussion on learning data at the end.

Here is the keynote video, which you can also find on Rod’s Pulse Podcast page. Slides are available at http://bit.ly/el3hill

Welcome to the PhilOnEdTech blog

, /by

After nearly 8 years co-publishing the e-Literate blog, and after nearly 7 years in business at MindWires with Michael Feldstein, we  are moving on. Michael is leaving MindWires and focusing his efforts on the Empirical Educator Project with e-Literate closely aligned. We plan to accelerate our consulting work at MindWires, our market analysis work with our partners at LISTedTECH (currently the LMS Market Analysis, but we’ll roll out more starting this summer), and I will be using this blog as my new writing home. You should see continued market analysis and commentary from me as well as from a curated group of guest bloggers.

If you were subscribed for newsletter updates at e-Literate, you should already be subscribed for PhilOnEdTech. Expect to see a confirmation email and ability to opt out if you do not wish to get both newsletters. If you have not yet subscribed, you can easily add your name and email through the widgets on the side bar and in the footer.

I have thoroughly enjoyed working and writing with Michael, and I wish him the best of luck. You should follow the great work he’s leading at Empirical Educator Project and at e-Literate. I’ll miss working with him, but I am also excited for both of us to be able to focus more on our individual passions.

Christensen Scorecard: Data visualization of US postsecondary institution closures and mergers

, /by

In 2013, Harvard Business professor Clayton Christensen made a bold prediction based on his ubiquitous innovation theory that maybe half of all postsecondary institutions could close within 10-15 years.

(source: https://youtu.be/KYVdf5xyD8I, starting at 6:25)

The scary thing is that 15 years from now, maybe half of the universities will be in bankruptcy, including the state schools. But in the end, I’m excited to see that happen.

Christensen then doubled down on his predictions in 2017, humorously saying it might take nine years instead of ten.

(source: https://youtu.be/4ljlUOV-Uj4, starting at 1:04:42)

Q. Do you still believe, as you’ve said before, that as many as half of colleges and universities will be bankrupt or closed within a decade?

A. Um, yes. [snip] Whether the providers get disrupted within a decade — I might bet that it takes nine years rather than 10. Maybe I’m too scared about the Harvard Business School to be rational about it. But we should worry.

There have been plenty of articles written about these claims, but it has been frustrating that very few back up their analysis with data. One exception is Derek Newton’s article critiquing the claims in Forbes, titled “No, Half Of All Colleges Will Not Go Bankrupt”.

Look at the numbers. In the 2013-14 year, there were 3,122 four-year colleges according to the Department of Education. In 2017-18, the most recent data, there were 2,902 – a drop of about 7% over four years. That could be disruptive. But numerically, all of school closures since Christensen made his 2013 forecast were four-year, for-profit schools, which fell from 769 in 2013 to 499 in 2017 – a drop of 270. Of all the colleges, at all levels, that have closed since 2013, 95.5% of them were for-profit institutions.

Another exception is Michael Horn’s explanation of the predictions (he co-authored the New York Times op-ed from 2013, titled “Innovation Imperative: Change Everything”, that included the initial prediction). This 2018 post “Will half of all colleges really close in the next decade?” also sought to go back to original, more nuanced claims of 25% closures and mergers at the Christensen Institute.

Translation? Our predictions may be off, but they are directionally correct.

To that I emphasize one more piece of nuance. Ultimately we are really predicting a failure rate, made up of a combination of closures, mergers or acquisitions, and bankruptcies in which a college or university has the opportunity to restructure itself. Not all universities that “fail” will disappear. [snip]

From 2004–2014, “Closures among four-year public and private not-for-profit colleges averaged five per year from 2004-14, while mergers averaged two to three,” according to Moody’s. Moody’s predicted in 2015 that that closure rate—out of 2,300 institutions—would triple by 2017, and the merger rate would double.

Assuming that were true, and say that the rate held steady for 15 years, that would take out roughly 13% of existing higher education institutions right there.

Thanks to our partners with our LMS Market Analysis service, LISTedTECH, we can now provide data visualizations to better evaluate the validity or likelihood of these claims. For the first time that I’m aware of, we have visualizations showing combined closures and mergers over time, broken down by sector and degree-type, and showing data 2-3 years in advance of IPEDS publications.

The LISTedTECH data shown below tracks known closures and mergers, which have then been checked against both IPEDS and Federal Student Aid data sets. There are translation issues in all three data sets, so the data will not match 100% – probably more at the 80 – 90% confidence level. The first view shows combined closures and mergers per year, broken out by control and whether they are classified as 2-year or 4-year degree-granting institutions.

Closed US higher ed schools over past decade

As Derek Newton and Michael Horn pointed out, the vast majority of closures were from the for-profit sectors. Part of the dynamic at play is that when a large for-profit chain meets its demise (e.g. Corinthian Colleges, ITT, Westwood Colleges) or has a massive downturn (e.g. University of Phoenix) literally dozens of individual institutions close, whereas when a small private nonprofit college in New England closes, it is one school. Add to the that the massive drop in for-profit enrollments since 2012.

The public sector data in 2013 and 2014 is largely driven by reorganizations in the University System of Georgia.

Also note that the 2019 data only includes the first quarter.

If we want to track the Christensen (and Horn) predictions, however, we need to view this data as a running total.

Running total of closed and merged US higher ed institutions

Let’s zoom out to capture the timeline of the most recent predictions of a decade from 2017, and let’s add the rough levels indicated (using bold row from this IPEDS table to define number of institutions).

Running total of closed US institutions with trend lines

If you include all degree-granting institutions (i.e. for-profits as well as private nonprofits and publics), then the current trends lines show that the 50% closure prediction by 2027 certainly seems feasible. Note, however, is that there are less than 1,000 for-profit institutions remaining as of Fall 2017 IPEDS data, and the rate of for-profit closures cannot continue more than another 8-10 years (best case / worst case, take your pick).

There are quite a few stories recently about private nonprofit small-school closures, but the data thus far don’t show a rapid acceleration of closures. Some perspective is useful here.

If you ignore the for-profit sectors, then the trend line for private nonprofit and public institution closures + mergers remains far below that needed to hit the 25% level described by Horn or the 50% level described by Christensen. None of this is to say that the trends moving forward will be linear, however. The rate of private nonprofit and public closures and mergers would need to at least triple to hit the more conservative level of 25% within a decade, a possibility that I would not reject out of hand. And it turns out that Moody’s was wrong – the rate of closures and mergers in this group did not triple from 2015 – 2017. Nevertheless, the data could get worse.

We’ll share more information on this new data, but hopefully these visualizations provide a better sense of the trends on college closures and mergers.