Entrepreneurship and Mentorship in Online Courses

Entrepreneurship Research on NovoEd

Chuck EesleyProfessor, Management Science &

EngineeringStanford University

Why I’m Using NovoEd Data

• Ability to easily run randomized experiments and infer causality in the entrepreneurship process is unprecedented.

• “Real world” students/startups rather than undergrads

• Large N– ~25k students*6 iterations of the class = >150k students – *every interaction on the platform = millions of

datapoints

Current Interest: Mentorship

• Mentorship programs are increasingly on the agenda for policymakers and universities interested in fostering entrepreneurship.

– Few studies examine causal effects of mentorship

– 2 studies, 1 on NovoEd, one with Stanford E145 class

Stanford Study• We investigate the impact of the type of mentorship on the

likelihood that university students will become entrepreneurs.

• We use a longitudinal field experiment with a pre-test/post-test design where students in an entrepreneurship class were randomly assigned to receive mentorship from either entrepreneur or non-entrepreneur mentors.

• To our knowledge, this is the first randomized trial of a mentoring program in entrepreneurship. We find significant positive effects of mentorship, particularly by certain types of mentors.

2 3 4 5 Overall0

0.1

0.2

0.3

0.4

0.5

0.6

Without entre-preneur mentorWith en-treperneur men-torBoth groups

Family: 0 Family: 1 Overall0

0.1

0.2

0.3

0.4

0.5

Without entrepreneur mentorWith entreperneur mentorBoth groups

0.2

.4.6

.81

Pr(

Pos

t-cla

ss s

tartu

p ex

perie

nce)

2 3 4 5Willingness to take risk

Ent_mentor=0 Ent_mentor=1

0.1

.2.3

.4.5

.6P

r(P

ost-c

lass

sta

rtup

expe

rienc

e)

0 1Family background in entrepreneurship

Ent_mentor=0 Ent_mentor=1

Coming from families with little entrepreneurial experience, teaming students with entrepreneur mentors dramatically increases their likelihood of embracing a startup career (i.e. from around 12% to 22%, or about 45.5% in relative magnitude).

Table 3: Logit Regressions for the Relationship between Entrepreneur Mentor and Students’ Choice of Startup Careers

Results• Results show that entrepreneur mentors have a significant positive

influence on the rate of entrepreneurship, with the greatest influence on students with specific risk orientation and family backgrounds.

• However, this study took 4 years and resulted in a final sample size of only 153. (in contrast NovoEd study has 25k students every 7 week course)

• Eesley, Charles E. and Wang, Yanbo, The Effects of Mentoring in Entrepreneurial Career Choice (January 29, 2014). Available at SSRN: http://ssrn.com/abstract=2387329 or http://dx.doi.org/10.2139/ssrn.2387329

NovoEd study

Using data from >150,000 students

Students were randomized to one of6 different versions of the mentorassignment.

•2 different types of mentors•2 different approaches to workingw/ a mentor most effectively.

Outcomes:•Engagement on the platform•Scores on the assignments•Continuing with the startup after class•Successful fundraising/product•release after class

• Group A (Control Group): Just ask students to find a mentor;

• Group B: Find a diverse connection mentor using predictive logic;

• Group C: Find a similar connection mentor using predictive logic;

• Group D: Find a diverse connection mentor through being flexible and non-predictive logic;

• Group E: Find a similar connection mentor through being flexible and non-predictive logic

• In addition to randomizing on the type of assignments, we could also randomize on the course content for the students.

• In the future, content can vary based on the 5 groups as listed above and we could teach teach each group different strategies to finding a mentor and becoming an entrepreneur.

• A next study under consideration is randomization in team formation.

NovoEd study

NovoEd data

• DVs– Engagement on the platform, forums, etc.– Scores on each assignment, feedback– End of class survey (also after class surveys)

• IVs and control variables– Individual characteristics– Team characteristics– Mentor characteristics– Version of mentorship assignment (randomized)

Descriptive statistics

Goals

• We hope to use the NovoEd platform to develop a methodology where various social sciences theories can be easily implemented and tested.

• The results from this process can be used for academic publication and also in training new entrepreneurship faculty and lecturers to share best practices in a data-driven way.