The MOOC pivot

Abstract

When massive open online courses (MOOCs) first captured global attention in 2012, advocates imagined a disruptive transformation in postsecondary education. Video lectures from the world's best professors could be broadcast to the farthest reaches of the networked world, and students could demonstrate proficiency using innovative computer-graded assessments, even in places with limited access to traditional education. But after promising a reordering of higher education, we see the field instead coalescing around a different, much older business model: helping universities outsource their online master's degrees for professionals. To better understand the reasons for this shift, we highlight three patterns emerging from data on MOOCs provided by Harvard University and Massachusetts Institute of Technology (MIT) via the edX platform: The vast majority of MOOC learners never return after their first year, the growth in MOOC participation has been concentrated almost entirely in the world's most affluent countries, and the bane of MOOCs—low completion rates—has not improved over 6 years.

Full text

This manuscript is a post-print version of the document published in:
Reich, J. and Ruipérez-Valiente, J. A. (2019). The MOOC Pivot.
Science 363(6423), pages 130-131.
DOI: 10.1126/science.aav7958
http://science.sciencemag.org/content/363/6423/130
© 2019 Science/AAAS Publications

Post-print Manuscript.
Original: http://science.sciencemag.org/content/363/6423/130
1
The MOOC Pivot: From Teaching the World to Online Professional Degrees
Authors: Justin Reich1, José A. Ruipérez-Valiente1
Affiliations:
1Massachusetts Institute of Technology.
Abstract: When massive open online courses (MOOCs) first captured global attention in 2012,
5
advocates imagined a disruptive transformation in postsecondary education. Video lectures from
the world's best professors could be broadcast to the farthest reaches of the networked world, and
students could demonstrate proficiency using innovative computer-graded assessments, even in
places with limited access to traditional education. But after promising a reordering of higher
education, we see the field instead coalescing around a different, much older business model:
10
helping universities outsource their online master's degrees for professionals. To better
understand the reasons for this shift, we highlight three patterns emerging from data on MOOCs
provided by Harvard University and Massachusetts Institute of Technology (MIT) via the edX
platform: The vast majority of MOOC learners never return after their first year, the growth in
MOOC participation has been concentrated almost entirely in the world's most affluent countries,
15
and the bane of MOOCs—low completion rates—has not improved over 6 years.
One Sentence Summary: MOOCs pivot from teaching the world to online professional degrees.
Main Text: In 2012, advocates for massive open online courses (MOOCs) imagined a disruptive
transformation in post-secondary education. Video lectures from the world’s best professors
could be broadcast to the farthest reaches of the networked world, students could demonstrate
20
proficiency using innovative computer-graded assessments, and new global onramps could
expand access to education. Coursera, edX, and later FutureLearn attempted to build online
course catalogs that reflected the full disciplinary breadth of universities, from STEM subjects,
to the humanities, to the professions (1). Coursera co-founder Daphne Koller described their
business model as a “blue ocean strategy” (2): they would sustain a new global service by
25
converting non-consumers of higher education—especially in places with limited access—into
online learners at the world’s best universities (3,4,5). MOOC providers would make learning
materials freely and widely available, and they would earn revenue from a portion of learners
who purchased the opportunity to earn verified certificates and credentials.
That didn’t work at Coursera or any other MOOC provider, and the field is now coalescing
30
around a well-established business model: helping universities outsource instruction, technology,
and marketing for online Masters degrees.
In October of 2018, edX became the last of the major MOOC providers to announce partnerships
with universities to offer fully online professional Masters degrees (6), five years after Udacity
made the first such partnership with Georgia Tech. EdX’s move into online professional
35
education follows their May 2018 decision—mirroring earlier decisions by Coursera and
Udacity—to begin building paywalls around their previously freely available content (7). Rather
than expanding free learning around the world, MOOC providers will now compete with well-
established for-profit companies in helping universities outsource their online degrees.

Post-print Manuscript.
Original: http://science.sciencemag.org/content/363/6423/130
2
For two decades, a class of companies called “online program managers” or “school-as-a-
service” companies—Pearson Embanet, 2U, Wiley Education Services—have supported
colleges in creating online degrees (8). These school-as-a-service providers offer services
ranging from marketing and recruitment, admissions, online course management, curriculum
design, or course instruction and assessment, and then universities choose how much of the total
5
student experience to outsource to these providers. School-as-a-service providers typically earn
revenue by taking a fraction of the tuition of each student enrolled.
MOOC providers are reorienting their efforts to compete directly with these existing companies
in one market segment: professional Masters degrees, credentialed by near-top universities, in
fields with well-established “return on investment”: business, computer programming, data
10
science, and related fields. The primary competitive advantage of MOOCs relative to established
school-as-a-service providers involves cutting labor costs through automation. Many
“traditional” online programs include small class sizes, synchronous sessions with instructors,
and human-graded assignments. Degrees offered by universities with the technology and support
of Coursera and edX will be one-half or one-quarter as expensive as typical American
15
professional online credentials, with the bulk of savings coming from a combination of larger
class sizes, fewer or no synchronous sessions, reduced contact with instructors, and more
autograded assignments (9).
After six years of heralding themselves as a disruptive force in education, why would MOOC
providers find themselves competing directly in a well-established service on familiar terms with
20
for-profit publishers and consultants?
Typically, the operations of for-profit school-as-a-service providers are opaque. Their business
model involves educational institutions outsourcing their core competencies to for-profit
operators, so both online program managers and universities are usually eager to avoid scrutiny.
In the case of edX, however, the founding partners—Harvard and MIT—make data about their
25
MOOC courses available to researchers, so we can investigate patterns of global participation in
HarvardX and MITx courses over the past six years to better understand the challenges of the
“blue ocean strategy” and the reasons for focusing on professional Masters degrees.
The data that we analyze in this paper comes from all MITx and HarvardX MOOCs taught on
edX from the start of the initiative to May 2018. The dataset includes 565 course iterations from
30
261 different courses with a combined 12.67 million course registrations from 5.63 million
learners. Data from other edX partners or MOOC providers might reveal different dynamics, but
we have a detailed view of two of the largest course providers. In reviewing these data, three
patterns are newly salient: the vast majority of MOOC learners never return after their first year,
the growth in MOOC participation has been concentrated almost entirely in the world’s most
35
affluent countries, and the bane of MOOCs—low completion rates (10)—have not improved
over six years.
MOOC researchers realized early on that most MOOC registrants leave soon after enrollment. Of
those who register for a course, 52% never enter the courseware (Table S4), and attrition
typically remains high in the first two weeks of a course (10). We see similar patterns looking at
40
engagement over multiple years. In Figure 1, we show the year-to-year enrollment of learner

Post-print Manuscript.
Original: http://science.sciencemag.org/content/363/6423/130
3
cohorts defined by their year of first activity. New unique learners increase for the first four
years but have declined since. The largest initial cohort was in year 4 (2015-2016) with 1.1
million unique learners, but only 12% of those learners took an additional HarvardX MITx
course in the following year. Cohorts since 2015-2016 have been half the size of largest cohort.
Second year retention rates have declined with every cohort, from 38% in the first cohort to 7%
5
in the year 5 (2016-2017) cohort. A growing global demand for ongoing learning from MOOCs
that might have maintained a blue ocean strategy never materialized.
Figure 1. Churn rate of unique learners by registration year. Parenthesis shows percentage
10
retained from initial cohort size.
It was clear from the first few years of MOOC research that MOOCs disproportionately drew
their learners from affluent countries and neighborhoods, and markers of socioeconomic status
were correlated with greater persistence and certification (11,12). With six years of data, an
15
additional trend comes into view: nearly all of the growth (and subsequent decline) in new
registrations and certifications came from the world’s most affluent countries. In Figure 2, we
show the number of registrants and certificates per year divided into quartiles based on the UN
Human Development Index (HDI) rating of each registrant’s home country. Rather than creating
new pathways at the margins of global higher education, MOOCs are primarily a complementary
20
asset for learners within existing

Post-print Manuscript.
Original: http://science.sciencemag.org/content/363/6423/130
4
systems.
Figure 2. Number of enrolments and certifications by year separated by Human
Development Index category.
5
Finally, the bugaboo of MOOCs, their low completion rate (10), has not budged in six years. In
Figure 3, we show completion rates for three non-exclusive groups: the rate for all course
participants, for all learners who indicate on a survey that they intend to complete a course, and
for all learners who pay for a verified track. Six years of investment in course development and
learning research has not produced meaningful improvements in these figures (13, 14). A blue
10
ocean strategy that depends on bringing new learners into higher education cannot succeed if
educational institutions cannot support learners in converting their time and financial investment
into a credential with labor market value.

Post-print Manuscript.
Original: http://science.sciencemag.org/content/363/6423/130
5
Figure 3. Percentage of course completion by year and cohort of learners.
If MOOC growth is stagnant, if most learners never return after signing up for a course, if
growth is concentrated in the wealthiest parts of the world, and if course completion remains
5
relatively low even among those paying for courses, then financial sustainability for MOOC
platforms may depend upon reaching smaller numbers of people with greater financial means
that are already embedded in higher education systems rather than bringing in new non-
consumers from the margins. MOOC providers are on a pathway to focus on already-affluent,
already-educated learners in the developed world who can afford a $21,000 online Masters in
10
Accounting from Indiana or a $22,000 online MBA from Illinois. After promising to disrupt
higher education, MOOC providers may settle for cutting costs through automation.
For universities that joined the MOOC movement with a mission-driven goal of expanding free
access to education, those faculty and leaders will need to revisit whether the current direction of
MOOC platforms is aligned with that mission.
15
As MOOC platforms support programs that look more like “traditional” online higher education,
the literature on online learning can provide useful guidance. By most indications, students
typically do worse in online courses than in residential courses, and the challenges of online
learning are particularly acute for the most vulnerable populations of first-generation college
students, students from low-income families, and under-represented minorities (15). If low-cost,
20
MOOC-based degrees end up recruiting the kinds of students who have historically been poorly
served by online degree programs, student support programs will be vital. Some recent research
has explored online and text-message-based interventions for supporting these students (16), but

Post-print Manuscript.
Original: http://science.sciencemag.org/content/363/6423/130
6
most established research suggests that human connections through advisors, tutors, and peer
groups provide the most important student supports (17, 18), and these human supports will push
against lower tuition costs. MOOC-based degree providers may find that highly effective online
learning for diverse populations costs about the same to provision as highly effective residential
learning (9).
5
MOOCs will not transform higher education, and they probably will not disappear entirely.
Rather, they will provide new supports for specific niches within already existing education
systems, primarily supporting already-educated learners. The six-year saga of MOOCs provides
a useful cautionary tale and framing device for education policymakers facing down whatever
will be the next promoted innovation in education technology, be it artificial intelligence or
10
virtual reality or some unexpected new entrant. New education technologies are rarely, perhaps
never, disruptive; rather they are domesticated by existing cultures and systems (19). Dramatic
expansion of educational opportunities to underserved populations will require political
movements that change the focus, funding, and purpose of higher education; they will not be
achieved through new technologies alone.
15
References and Notes:
1. Chuang, A. Ho. HarvardX and MITx: Four years of open online courses--fall 2012-
summer 2016 (2016).
2. W. C. Kim, R. Mauborgne, California management review 47, 105-121 (2005).
20
3. D. Koller, A. Ng. The online revolution: Learning without limits. In Proceedings of the
20th International Conference for Online Learning (2013). Review retrieved from
https://sloanc2013.wordpress.com/2013/11/21/moocs-the-blue-ocean-strategy/ ; Slides
from similarly named talk retrieved from
https://www.wiche.edu/info/walf/meetings/annual2013/meetingMaterials/kollerOnlineRe
25
volution.pdf.
4. M. Trucano. MOOCs in Africa. EduTech: A World Bank Blog on ICT Use in Education.
(http://blogs.worldbank.org/edutech/moocs-in-africa).
5. Terwiesch. Introduction to Operations Management: Conclusion
(https://www.youtube.com/watch?v=igEx_DQ4D5c&feature=youtu.be)
30
6. Agarwal. Fully Online, Top-Ranked Master’s Degrees Now Available on edX. edX blog
(https://blog.edx.org/fully-online-top-ranked-masters-degrees-now-available-edx, 2018).
7. A. Agarwal. Furthering the edX Mission, Forging a Future Path. edX blog
(https://blog.edx.org/furthering-the-edx-mission, 2018).
8. P. Hill, Educause Review 47, PP (2012).
35
9. P. Hill. Coursera CEO Interview: Betting on OPM market and shift to low-cost masters
degrees. E-literate (https://mfeldstein.com/coursera-ceo-interview-betting-on-opm-
market-and-shift-to-low-cost-masters-degrees/)
10. J. Reich. MOOC Completion and Retention in the Context of Student Intent. Educause
Review. (https://er.educause.edu/articles/2014/12/mooc-completion-and-retention-in-the-
40
context-of-student-intent)
11. J. D. Hansen, J. Reich, Science 350, 1245-1248 (2015).
12. R. F. Kizilcec, A. J. Saltarelli, J. Reich, G. L. Cohen, Science 355, 251-252 (2017).
13. J Reich. Science. 347, 34-35 (2017)
14. J Reich. NAE Bridge. 46(3) 29-37 (2016)
45

Post-print Manuscript.
Original: http://science.sciencemag.org/content/363/6423/130
7
15. D. Xu, S. S. Jaggars, The Journal of Higher Education 85, 633-659 (2014).
16. B. Castleman, Behavioral Insights for Federal Higher Education Policy. Urban Institute
(2017).
17. S. Scrivener, M. Weiss, A. Ratledge, T. Rudd, C. Sommo, H. Fresques. Doubling
graduation rates: Three-year effects of CUNY's Accelerated Study in Associate Programs
5
(ASAP) for developmental education students. MDRC (2015).
18. D. Xu, S. Solanki, P. McPartlan, B. Sato, Educational Researcher 47, 435-450.
0013189X18778559 (2018).
19. L. Cuban. Teachers and Machines. Teachers College Press. (1986)
20. G. Lopez, D.T. Seaton, A. Ang, D. Tingley, I. Chuang. Google BigQuery for Education:
10
Framework for Parsing and Analyzing edX MOOC Data. Proceedings of the Fourth
ACM Conference on Learning@Scale, 181-184. (2017).
21. United Nations. Human Development Indices and Indicators: 2018 Statistical Update.
http://hdr.undp.org/sites/default/files/hdr2018_technical_notes.pdf (2018)
15
Acknowledgments: Funding: No specific funding sources supporting this essay; Author
contributions: Reich conceptualized the paper, wrote the original draft, and revised. Ruiperez-
Valiente conceptualized the paper, conducted data analysis, and revised the paper. Competing
interests: Authors declare no competing interests; Data and materials availability: MITx and
HarvardX learner data are treated as student data meriting FERPA protections. MITx data can be
20
requested at http://web.mit.edu/ir/mitx/ and HarvardX data at https://vpal.harvard.edu/vpal-
research-request.
Supplementary Materials:
Materials and Methods
Tables S1-S4 and Figure S1
25

1
Supplementary Materials for
The MOOC Pivot: From Teaching the World to Online Professional Degrees
Justin Reich, José A. Ruipérez-Valiente
Correspondence to: jreich@mit.edu
This PDF file includes:
Materials and Methods
Tables S1 to S4
Fig. S1

2
Materials and Methods
This supplementary section describes the materials and methods that have been used to
produce the research and how to access instructions to reproduce the analysis presented. If you
have any additional questions, do not hesitate to query directly the authors of this work.
Context and Data Collection
EdX was created in May 2012 by faculty from MIT and Harvard universities with the objective
of expanding education from the best universities in the world worldwide. EdX currently have
more than 100 partners that are producing and teaching courses on their platform. Additionally,
the software known as Open edX that runs the platform is open source and has become a global
collaborative initiative used by multiple institutions around the world.
The data collection that we analyze in this study comes from all the MOOCs taught on edX by
the founding partners MITx and HarvardX from the start of the initiative to May 2018. An
overview of the dataset includes 565 course iterations from 261 different courses that have
summed more than 12.67 million course registrations from over 5.63 million learners that
generated over 4.4 billion events and invested more than 48 million hours in such courses. We
organize these courses into annual cohorts running from June to May. (Note that in some
previous HarvardX/MITx research, annual cohorts are organized from September to August.)
EdX collects all the click-stream data from their learners which them we process using
edx2bigquery (20) framework to build up a person-course dataset with over 60 variables that
describe the interaction of the student with every course they enrolled to. We also use the Human
Development Index (HDI) and Category provided by the United Nations (UN) for 2017 (see
http://hdr.undp.org/en/content/human-development-index-hdi).
Variables:
HDI: It is a country composite measure provided by the UN based on three dimensions:
health, knowledge and general quality of living.
Course enrollments: Number of learners that registered for a course.
Unique learners: Number of different learners independently of how many course
enrollments they did.
Participants: Learners that registered and viewed the course.
Explorers: Learners that viewed at least half of the chapters of a course.
Completers: Learners that achieved a passing grade at the end of a course.
Verified: Learners that paid to enter the verified certificate track.
Certified: Learners that achieve a passing grade and received a verified certificate.
Registration year cohort: We create a cohort for each year with the learners’ that did the
first registration for a course during that year.
Percentage retention: Percentage of learners from a registration year cohort that registered
for a course during that year.
Intention to complete: Self-reported measure as part of the pre-course survey regarding
students’ intention to complete the course.

3
Reproducing the analysis:
We have prepared a GitHub repository (https://github.com/jruiperezv/MOOC_Pivot) with
instructions on to reproduce the analysis that contains:
1. How to access the data: MITx and HarvardX learner data are treated as student data
meriting FERPA protections and thus cannot be released publicly. MITx data can be
requested at http://web.mit.edu/ir/mitx/ and HarvardX data at
https://vpal.harvard.edu/vpal-research-request.
2. Specific dataset, fields and courses needed to reproduce the analysis.
3. Additional data needed with course and country metadata.
4. Description of all data fields.
5. IPython notebook with the code to reproduce the data from tables S1-S4 presented in
these supplementary materials which are the basis of the analysis presented in this study.
6. RMarkdown script to reproduce raw visualizations of Figs. 1 to 3 based on the csv files.

4
Table S1.
Aggregate data of Figure 1: Churn rate of unique learners by registration year.
Registration
year cohort
Year
Unique
learners
Percentage retention
2012-13
2012-13
192255
100
2012-13
2013-14
72093
37.5
2012-13
2014-15
46494
24.18
2012-13
2015-16
36089
18.77
2012-13
2016-17
21690
11.28
2012-13
2017-18
15059
7.83
2013-14
2013-14
755800
100
2013-14
2014-15
205814
27.23
2013-14
2015-16
146713
19.41
2013-14
2016-17
76780
10.16
2013-14
2017-18
51996
6.88
2014-15
2014-15
630992
100
2014-15
2015-16
145499
23.06
2014-15
2016-17
60971
9.66
2014-15
2017-18
38530
6.11
2015-16
2015-16
1108858
100
2015-16
2016-17
129562
11.68
2015-16
2017-18
92977
8.38
2016-17
2016-17
531213
100
2016-17
2017-18
53030
9.98
2017-18
2017-18
499736
100

5
Table S2.
Aggregate data of Figure 2. Number of course enrollments and certifications by year separated
by HDI category.
Year
HDI Category of
Human Development
Number of
enrolments
Number of
certifications
Percentage
enrolments
within year
Percentage
certified
within year
2012-13
Very high
205955
10798
59.28
61.98
2012-13
High
54483
2476
15.68
14.21
2012-13
Medium
74395
3351
21.41
19.24
2012-13
Low
12616
796
3.63
4.57
2013-14
Very high
1359191
43235
65.55
71.55
2013-14
High
318683
8429
15.37
13.95
2013-14
Medium
357971
7685
17.26
12.72
2013-14
Low
37780
1075
1.82
1.78
2014-15
Very high
1583635
55674
62.82
71.2
2014-15
High
414912
10903
16.46
13.94
2014-15
Medium
474704
10373
18.83
13.27
2014-15
Low
47543
1244
1.89
1.59
2015-16
Very high
1798426
46790
61.95
71.66
2015-16
High
483046
9659
16.64
14.79
2015-16
Medium
569422
8174
19.62
12.52
2015-16
Low
52073
675
1.79
1.03
2016-17
Very high
947969
17827
57.44
68.71
2016-17
High
335067
4626
20.3
17.83
2016-17
Medium
329609
3056
19.97
11.78
2016-17
Low
37632
436
2.28
1.68
2017-18
Very high
954426
14341
55.58
68.7
2017-18
High
305950
3325
17.82
15.93
2017-18
Medium
401982
2912
23.41
13.95
2017-18
Low
54975
298
3.2
1.43

6
Table S3.
Aggregate data of Figure 3. Percentage of course completion by year and cohort of learners. The
first cohort represents all participants. All cohorts are participants, the second cohort is learners
that report intention to complete and the third cohort are learners that paid to enter the verified
track. Cohorts two and three are not mutually exclusive.
Year
Cohort
Size of cohort
Number of
completions
Percentage completion
2013-14
Participants
1219030
60285
4.96
2014-15
Participants
1333780
78085
5.91
2015-16
Participants
1958222
80495
4.13
2016-17
Participants
1097759
43726
3.98
2017-18
Participants
980589
30650
3.13
2013-14
Intention to
complete
178909
20593
11.51
2014-15
Intention to
complete
191578
33986
17.74
2015-16
Intention to
complete
209333
34416
16.44
2016-17
Intention to
complete
74116
11952
16.13
2017-18
Intention to
complete
42005
6478
15.42
2013-14
Verified
6790
3003
44.23
2014-15
Verified
27490
13792
50.17
2015-16
Verified
52384
26051
49.73
2016-17
Verified
47362
26725
56.43
2017-18
Verified
46326
21321
46.02

7
Table S4.
Mean, median and standard deviation of the percentage of participants, explorers and certified by
course (N = 565).
Year
Mean
Median
Std. Deviation
Percentage participants
48.02
51.01
15.49
Percentage explorers
25.59
19.84
19.54
Percentage certified
5.03
2.88
6.04

8
Fig. S2.
Boxplot visualization of the percentage of participants, explorers and certified by course.