Across the world, outreach professionals are finding that text messaging can be an effective platform to address social issues by prompting changes to individuals' behavior and serving as an outlet to provide personalized counseling and support. This presentation focuses on the experiences of two organizations - the West Virginia Higher Education Policy Commission (HEPC) and uAspire - in utilizing text messaging to promote college access and support college retention and success. The projects featured are supported by the work of Signal Vine, LLC, a company specializing in personalized, two-way text messaging platforms tailored to the needs of education organizations. HEPC is a state agency which serves as the coordinating body for West Virginia's public four-year colleges and universities. uAspire is a non-profit organization focusing on college affordability. This work is informed by the research of Drs. Ben Castleman and Lindsay Page and funded in part by the Kresge Foundation. Co-authors/presenters: Dr. Sarah Beasley, Alexandra Chewning, and Brian Kathman.
2. Road map
Day 2:
• Needs assessment
• Technology
• Messaging
• Implementation
Day 1:
• Introductions
• Research Review
• Case Studies
3. Introductions
Let’s Get to Know Each Other!
• Name
• Department/Organization & Role/Title
• Starburst Question
• One thing you hope to learn
4. Research review
• Prior studies and trials
• Principles of behavioral economics
• Other policy interventions
• Key publications and resources
5. Prior studies and trials
Session informed by leading researchers in this
field - Dr. Ben Castleman and Dr. Lindsay Page
● Use principles from behavioral economics
& psychology to improve college outcomes
● Focus their work on supporting low-income
& non-traditional students
● Powerful, exciting results have garnered
national attention & widespread application
6. Prior studies and trials
2010: uAspire began partnership
with Castleman & Page to study
summer melt 20% drop-off
2011: First summer melt
intervention designed - successful
but difficult to reach students
Summer Melt: the phenomenon whereby
college-intending students do not make the
transition from high school to college
7. Prior studies and trials
• Advisors reached out to HS graduates
in summer to support their transition
• Statistically significant impact on
enrollment - more students went to
college in the fall!
• One year later - even greater
statistically significant impact on
persistence
First summer melt study (2011)
8. Prior studies and trials
First texting study (2012): Applied key
ingredients of summer melt intervention to texting
9. Text messaging statistically
increased on-time college
enrollment by 7 percentage
points.
Among first-generation
students, there was an 8
percentage point increase in
enrollment in 4-year colleges.
2012 Text Messaging Results
Prior studies and trials
10. Prior studies and trials
Sent reminders to renew financial aid
and check SAP
Advisors texted back real-time
responses to student questions
Texting increased 2nd year persistence
by 12 percentage points
First College Texting Study (2013):
Restarted texting in January of freshman year
14. Key behavioral principles
Sources: Castleman (2013); Ideas42 (2014)
Limited attention
and planning fallacy
Adolescents are particularly prone to
put off hard choices/complex tasks in
favor of pleasurable pursuits — and
their attention often is divided.
15. Key behavioral principles
But adults also procrastinate when
faced with complexity.
Sources: Castleman (2013); Ideas42 (2014)
16. Key behavioral principles
Sources: Ideas42 (2014); Iyengar and Lepper (2000); Sivers (2009)
Status Quo Bias
• When overwhelmed, people
tend to do nothing.
• “Preset” or default options can
create monumental shifts in
behavior.
• “Choice is demotivating.”
17. Key behavioral principles
Sources: Ideas42 (2014); Castleman (2013)
Identity and Social Norms
• Individuals are highly influenced by their perceptions of
what’s normal for their peers.
• Individuals are motivated to meet clear expectations that
they deem to be standard or usual.
• The way we view and reflect on our personal identities or
roles has a powerful influence over our behaviors.
18. Key behavioral principles
Sources: Ideas42 (2014); Castleman (2013)
Coping with complexity
• Simplifying strategies
• Following the crowd
• Paralysis
19. Prior policy interventions
• Simplifying information increases participation in retirement
programs and the quality of school and college choices (Beshears et al,
2012; Hastings & Weinstein, 2008; Hoxby & Turner, 2013)
• Reducing hassles increases completion of federal financial aid
applications (Bettinger et al, 2012)
• Shifting perceptions of social norms increases tax compliance and
reduces home energy use (Alcott, 2011; Coleman, 1996)
• Prompts increase financial savings and flu vaccination rates (Karlan et al,
2010; Stockwell et al, 2012)
20. Prior policy interventions
• Simplifying information increases participation in retirement
programs and the quality of school and college choices (Beshears et al,
2012; Hastings & Weinstein, 2008; Hoxby & Turner, 2013)
• Reducing hassles increases completion of federal financial aid
applications (Bettinger et al, 2012)
• Shifting perceptions of social norms increases tax compliance and
reduces home energy use (Alcott, 2011; Coleman, 1996)
• Prompts increase financial savings and flu vaccination rates (Karlan et al,
2010; Stockwell et al, 2012)
21. Other interventions
• Weight Loss & Fitness
o Reify Health used text messaging in clinical trials
o Focused on weight-loss & fitness related interactions
• Ebola Outbreak
o The WFP is using texting to survey food supplies in Sierra Leone
(The Guardian, October 14, 2014)
• India Rural Farming
o Texting is used to communicate to rural villagers about subsidized
farming products
• Literacy in Niger
o World Bank Cell-Ed intervention increased basic literacy by 5 years
(World Bank, ‘Teach Literacy by Text. Really.”, 7/7/2014
22. Reading list
• Castleman and Page: Summer Nudging: Can Personalized Text Messages
and Peer Mentor Outreach Increase College Going Among Low-Income High
School Graduates (2013)
• Castleman: Prompts, Personalization, and Pay-offs: Strategies to Improve the
Design and Delivery of College and Financial Aid Information (2013)
• Ross et al: Using Behavioral Economics for Postsecondary Success (2013)
• Pew Research: Internet Project (see the myriad of reports related to mobile
internet access and use of text messaging among various populations)
• Castleman and Page: Summer Melt: Supporting Low-Income
Students Through the Transition to College (2014)
• Castleman and Page: Freshman Year Financial Aid Nudges
(2014)
26. Where is this happening?
• State education departments
–Louisiana, Arizona, Delaware, Montana, Massachusetts, Wisconsin, Virginia, and many
others in process
• US Department of Education
–Two funded research projects including an IES research grant and an Outward Bound
research study
–Inclusion in GEAR UP grant applications
•Major colleges and universities
–Research: Stanford, Harvard, Penn, U of Chicago, Michigan State, and Others
–Admissions & Student Affairs: U of Washington - Bothel, Youngstown State University, etc.
•College access orgs, school districts, and charters
–Philadelphia USD, KIPP Philly, Mastery, and many CAO’s
27. How is texting being used in Education?
• College-going process and summer melt
–College access organizations, K-12 districts
–To increase matriculation rates
• Admissions and financial aid
– Related to summer melt, improve engagement with students through their admittance
– Provide financial guidance up-front and renewal/compliance information throughout
– For financial aid, reduce loan default rates
•College persistence and student affairs
–Communicate with college freshman (Louisiana) to maintain good standing and persist
–Expanding to more general dialogue with students at the college/university level
• K-12 Attendance - (University of Chicago Research)
• Values Affirmation (Middle School) - (Stanford Research)
• Early Childhood Reading - (Stanford Research)
31. uAspire: College affordability non-profit
Impact in High-Need Cities:
● Direct Service to Students
○ Boston, Springfield, Lawrence &
Fall River, MA
○ Miami, FL
○ Bay Area, CA
● New Impact Models
○ Text messaging nationwide
Training & Technical Assistance:
● Training to Practitioners
○ School districts
○ Charter management orgs
○ Youth serving orgs
Policy & Advocacy:
● Represent student and family voice in
national dialogue
● Study & inform on affordability issues
32. Goals:
– Re-establish student-advisor contact
after high school graduation
– Nudge students on specific tasks to
enrollment: orientation, registration,
placement tests, financial aid & bill
– Provide customized advising content
using data from senior exit survey
uAspire texting: summer melt
33. uAspire texting: summer melt
Text messaging to support the high
school to college transition
2012: Piloted text messaging for
summer melt using online platform
• Students received 10 customized text
reminders July-August
• Could text message HELP but couldn’t
text back and forth
• Advisors called students in response
34. uAspire texting: summer melt
Identified need for a two-way, student-centered
texting platform.
2013: Redesigned platform so students could
interact “live” via text message
• Customized reminders sent out
• Students texted back for support
• Advisors responded immediately
35. uAspire texting: summer melt
Customized content based on student data:
Student City Data
Intended College Data Student City Data
36. uAspire texting in practice
Do students engage? Yes!
Real responses:
• I do need help. How do I file a loan?
• Yes, thank you! I have a lot of questions & the internet
is giving me little answers.
• I didn’t get accepted to any of the schools I applied to,
do I need to do FAFSA again for community college?
• Actually it’s great that you texted me bc I really do need
some help concerning my financial aid.
• OMG this came at just the right time. I need to
electronically SignMyLoan but I’m not sure how to do it.
37. But can advising happen via text? Yes!
uAspire texting in practice
38. uAspire evaluation and results
Used randomized design in each texting project:
-One group
randomly
assigned to
receive
texting
-One group
randomly assigned
to business as
usual
-More students than we could serve - lottery most equitable
-All students received support if they sought us out
-Allows for rigorous learning and lessons for the field
39. uAspire evaluation and results
Text messaging during
the summer statistically
increased on-time
college enrollment by 7
percentage points.
Among first-generation
students, we saw an 8
percentage point
increase in 4-year
enrollment.
Summer Melt Text Messaging Results
40. After summer melt success, uAspire expanded
texting into new domains:
1) Freshman year of college
2013-2014 - Reminders & real-time support for key
tasks: SAP, financial aid renewal, loan review
2) Parents of recent high school grads
Summer 2014 - Texted 650 parents about the
tasks required for their students’ enrollment
uAspire texting: expanded
41. uAspire texting: expanded
1) Freshman year text messaging results
● Text messaging into the freshman year of college
doubled the rate at which community college students
sought help from uAspire with their FAFSA renewal
● What’s more, freshman year
texting increased sophomore
persistence by 12 percentage
points among community
college students
43. Continued Summer Melt texting
• uAspire will bring texting program to our new Bay
Area site in summer 2015
Community College Pilot
• New project in collaboration with 3 large MA
community colleges to support 600 students via
texting and on-campus advising this school year
National Scale
• Will text juniors across several states beginning in
spring 2015
uAspire Texting: Next Steps
44. Building your audience
Opt-out
• Utilizes an existing
database to start
sending messages.
• Students can choose
to stop messaging
after an introductory
message is
sent/received.
Opt-in
• Students sign up to
start receiving text
messages (specifically).
• Sign-ups might include
written, digital or verbal
agreements.
vs
45. Behavioral Economics: Default
Austria (Opt-out)
99%
Source: Complete College America (2012). Guided pathways to success: Summary.
Germany (Opt-in)
12%
Organ
Donation
Rates
46. Opt-out model
When students have already agreed to
be contacted in some way…
Potential data sources:
• School administrative data
• Student surveys (online or paper)
• Scholarship applications
• College applications
• College registration paperwork
48. $225,000 grant from the
Kresge Foundation
Three-year pilot project
serving 14 high schools in
Year 1 (GEAR UP)
West Virginia’s project
49. Primary functions
Nudge
Urge students to complete
college-related tasks
Counsel
Provide access to one-on-
one counseling with a
professional
Align
Provide a continuous
contact point from high
school into college
50. Project goals
Financial aid awareness
• Increase the number of students
applying for and utilizing financial aid
51. Project goals
Campus connections
• Increase students’ use of campus
resources
• Position campus and state-level staff
as a continuous source of support
• Address “summer melt” by guiding
students through transitional process
52. Project goals
Capacity building
• Design a low-cost, high-impact service
model allowing colleges and
universities to improve access and
retention
• Expand on the body of research
surrounding next-generation student
supports and engagement
53. How it works
Aug - Dec:
students opt in
Jan: messages
begin
Mar: 1st college choice
prompt
Mar: campus
messages begin
May: students
graduate hs
May: 2nd college
choice prompt
July: final college
choice prompt
Aug: students enter
college
May: students finish
freshman year
June: messages
end
12th
grade
Year 2,
college
= students have direct access to a counselor at their college= 12th grade
= summer transition
= 1st year of college
54. How it works
Students opt in
• College admissions applications
• Statewide merit-based scholarship
application
• State-level college access web portal
• Inquiry forms collected at community
events and college fairs
55. Opt-in model
Potential data collection points:
• Applications
– College admissions
– Scholarship programs
• Events
– During the event
– During pre-registration
• Online
– Websites
– Social media pages
– Email blasts
• Text to join systems
57. How it works
Students receive scheduled messages
according to a pre-planned campaign.
Students can text our number at
any time for help.
A college counselor (a real person) responds!
58. How it works
State-level campaign:
• Messages are relevant to all
college-intending students.
• All students receive messages.
Example:
Did u know u can register 4 fall classes
now? Call 304-792-7098 & ask 4 a campus
counselor 2 make an appointment.
Example:
When reviewing your college schedule...
Remember '15 to finish.' On average, you have to
take 15 credit hours each semester to finish
college on time.
Campus campaigns:
• Messages provide information
specific to a particular college.
• Some students receive messages.
59. Six college partners:
• Bluefield State College (suburban public four-year, HBCU)
• Concord University (rural public four-year)
• New for 2015: Fairmont State University (suburban public
four-year)
• Marshall University (urban public four-year)
• Southern West Virginia Community and Technical
College (rural public two-year)
• New for 2015: West Virginia State University (urban
public four-year, HBCU)
How it works
60. How it works
Campus-level messaging
At various points in the campaign, students receive texts
asking if they’ve made a college selection…
• Students who select a partner college are added to
the campaign for that particular college.
• Students who select a non-partner college continue
receiving only state-level messaging.
• Students who do not respond or text back
“undecided” continue receiving state-level
messaging and are asked again about their college
choice later in the campaign.
61. HEPC commitments
• Provide text-messaging platform
• Provide technical support, training
and guidance
• Coordinate statewide messaging
62. Campus commitments
• Collect and share student data with
HEPC (name, cell number, email, high school,
and blind SSN)
• Align student supports to provide a
single point of contact
• Commit to following the statewide
schedule (with options of 2 custom
messages/month)
63. Campus commitments
• Provide staff to check messages
daily and within one hour of any
outgoing message
• Assist in the evaluation by providing
feedback and limited data
64. Campus staffing
Campus-level teams/support
Consider engaging…
• Admissions/recruitment officers
• Enrollment personnel
• Student affairs officers
• Student success/retention officers
• Academic/advising personnel
• Financial aid representatives
• Students (but only as mentors, not content experts!)
• Public information officers (keep them in the loop!)
66. Concord University
• Rural, regional public institution in southern WV
• ~2,900 enrollment (5-year average)
• 40% first-generation
• 53% Pell recipients
• 97% on financial aid
• 64% freshmen to sophomore retention rate
• 33% six-year graduation rate
67. Involve key stakeholders
Campus Team:
• Admissions, Financial Aid, Registrar,
Technology Services, Academic Success
Center, Advising, Student Affairs, etc.
• Student feedback
• Team lead and small team for refinement
68. What to include?
• What are important events, deadlines,
and information?
• How do we make students feel welcome
and excited about Concord (i.e. increased
sense of belonging)?
69. Responding to texts
• Reminder that text is going out
• Dedicated admissions and retention staff
members
• Timely, personalized responses
70. Next steps
• Expanding program
• Data analysis
• Utilizing text messaging in other areas
71. Early outcomes
1,000 +/- opt-ins
• Ten percent have since stopped
messaging (opted out).
Since January 2, 2014…
72. Early outcomes
1,800+/- student responses
• Questions about processes
Ex: When is the deadline to apply for WV financial aid?
• Requests for guidance
Ex: I can’t decide on a college... Help!
Since January 2, 2014…
77. • Data pollution due to student error
(e.g. students reporting the wrong grade level or
high school on sign up)
• Cell phone turnover (e.g. students change
numbers or pass the phone off to younger siblings)
• Lack of alignment with career and
technical pathways
• Fewer than expected responses
regarding college choice
• Declining engagement
Early challenges
78. Cell phone data can be
used to tailor Twitter
and Facebook
outreach
● Upload your cell phone list
to Twitter and/or
Facebook
● Automatically identify
individuals on your list for
targeting on these
platforms
● This works for email
addresses too!
Bonus feature
80. Evaluative study
Comparison groups:
• GEAR UP students who did not receive messages
• GEAR UP students who received messages
(segmented by length of participation… e.g. four months, 12 months)
• Non-GU students who did not receive messages
• Non-GU students who received messages
81. Evaluative study
Variables:
• Demographic (gender, race/ethnicity, Pell eligibility, zip code, high
school)
• Academic (ACT scores, GPA, highest high school math course
completed)
• Process (FAFSA completion, ACT/SAT score submission, payment of
deposits)
• Outcomes (Fall and spring enrollment, course completion, course grades,
degree objective, persistence)
90. Technology traps
“You wouldn’t just get on a loudspeaker”
One-Way Alerts
● A ‘Black Hole’ for communication
● No Support
● Reserve for Emergency Blasts
Download Required
● Will not reach everyone
● Very difficult to scale
● Several steps to lose students
Mass Text
● Same message for everyone
● Not relevant for many
● Quickly Ignored
Mobile Mass Marketing
● Impersonal communications
● Toeing the FERPA line
● Not process & outcome driven
91. The Signal Vine DNA
Relevance = Engagement = Results
2-way messaging
Provide students with direct
counseling; answer questions
Scheduling
Schedule messages as part of a
campaign, or for an individual
student
Personalization
Enable unique message content
and pathways
Case management
Responses are routed to the
appropriate counselor inbox;
Track histories and interactions
Message Intelligently
92. Why intelligent text messaging?
“It’s becoming the modern communication platform”
• Email is not working
–Only 3% of students check email daily
• Technology Advances
–Mobile phone advances put inexpensive “computers” in everyone’s hands
• Messaging is Preferred
–70%+ text daily with ‘Always On and Immediate’ communications
• Big Data
–Allows for mass personalization and optimization of messaging
• Scale
–Text allows for personalized interaction with thousands almost instantaneously
93. Text Messages Go Out
• “Program” of scheduled messages
• Personalized for each student
–Campus- and task- specific content & web links in messages
–Messages based on their data profile
94. Students reply
• Student Replies are routed to the appropriate Counselor Inbox
• Each Counselor Portal allows texting with many students at once
• Students reply when they need support
97. Demo: Setting the Stage
“Design a Custom Program of Messages”
• Time & Date
• Personalized Messages
• Conditions & Filters
–“Message Branches”
• Auto Replies
• “Listen” for Responses
– Save or Act
100. Other features
“It’s becoming the modern communication platform”
• Parent & Mentor Messaging
–Interact directly with parents and mentors
–Messaging can be driven by ‘Related Student Data’
• Multi-language Support
–Language type driven by student/parent/mentor profile
•Data Capture via Text
–Use to survey, get RSVP’s, ,etc.
103. Content and
messaging
• Anatomy of a text message
• Auto-responses, message pathways and
audience segmentation
• Style and tone
• Sample content
104. Best Practices
Credibility - Students need to recognize
and trust texting source
– Initial text is critical to establishing legitimacy
– Choose name carefully & explain program
– Personalize as much as possible
105. Best Practices
Responsiveness - Texting sets
expectation for immediate response
– Ensure texting schedule aligns with staff capacity
– Majority of text responses arrive within a few hours
– Let students know you’re standing by to respond
106. Best Practices
Clarity - Conveying information in
160 characters requires thoughtful
content & phrasing
– Pilot texts with students when possible
– Recognize confusion in student responses
& remedy immediately
– Elevate to phone conversation if needed
107. Best Practices
Frequency - Resist urge to text too
often or goals may be diluted
– Tell students how often you’ll be texting
– Text enough to keep momentum, not so much
that students tune out or become annoyed
– Be prepared for candid feedback
108. Best Practices
How often would you prefer to receive college planning text messages?
Source: survey of students who
attended West Virginia’s 2014 College
Goal Sunday event
109. Anatomy of a text
Element of personalization
Relevant information
Timeliness/urgency
Call to action/next step
Hi! It’s Jess w/ GEARUP. Did u know WV 4year colleges require either
the ACT or SAT? Need to take the ACT? Reg. by Jan 10 for the Feb test
http://bit.ly/1bUHmOs
110. Message pathways
Message personalization:
• Based on data fields (think MS mail merge!)
– Ex: Student first name
– Ex: Student’s advisor name
– Ex: Financial aid office phone number at the student’s
chosen college
Hi {first name}! Get free help completing your FAFSA during College
Goal Sunday on Feb. 15! There’s a location near you at {closest
College Goal Sunday site}. Register: http://samplelink.com
111. Message pathways
Audience segmenting:
• By campaign:
– Ex: FAFSA renewal campaign
– Ex: Junior year college-planning timeline
• By student variables:
– Ex: Students on academic probation
– Ex: Students who haven’t yet registered for orientation
• Management tool: By group or caseload:
– Ex: Marshall University students
– Ex: Scott High School students
112. Action automation
• Types of automation:
– Generate a message response
• Ex: “thanks” might trigger “you’re welcome!”
– Trigger a system action
• Ex: Stop messaging
• Ex: Move to a group or add to a campaign
• Ex: Launch message series
• Ex: Change future message schedule
113. Action automation
Example: When WV students say they’ve selected
a college, the system…
– Launches a series of response messages to gain more
information from the student.
– Automatically moves them to the appropriate caseload based on
their college choice.
– Automatically enrolls them in their particular college’s
supplementary messaging campaign.
– Automatically removes any further messages related to college
choice from their message schedule.
114. Style and tone considerations
Personality:
• Do you have a sense of humor?
• Are you informal or formal?
115. Style and tone considerations
Counseling approach:
• Proactive or reactive?
• How would you respond to this exchange?
CFWV: Don’t forget to register for orientation!
Student: I’m not going to college.
116. Style and tone considerations
Stop protocols:
• Honoring opt-outs
• Handling inappropriate messaging
117. Style and tone considerations
Brand and communications strategies:
• Style and use of terms
• Integration of organizational goals and
complementary campaigns
• Integration of partnerships
• Managing “hot button” issues and conflicts
123. External analytics
• Social views
– YouTube views
– Twitter follows
– Facebook likes
• Actions taken
– Call center calls
– Deposits or forms submitted
– Event registrations
• Surveys and polls
– How did you hear about us?
– Where do you get college planning information?
125. Policies and
procedures
• Legal requirements and responsibilities
• Partnership agreements
• Messaging and counseling guidelines
126. Legal requirements
• FCC rules prohibit SPAM and other
unwanted communications
– Nonprofit rules are more lenient than commercial
regulations
• Age limits
– COPPA restrictions require parental consent for
students under age 13
– School districts and other partners may have
individual policies
127. Legal requirements
• Student confidentiality and privacy
– FERPA regulations often apply
– Data sharing agreements should be in place
among partners and vendors
• Counseling policies
– Responsibilities to report
133. Contact
Sarah Beasley
Director of Retention
Concord University
(304) 384-6298
sbeasley@concord.edu
www.concord.edu
@campusbeautiful
Alexandra Chewning
Vice President of Research and Evaluation
uAspire
(617) 778-7195 x125
alexandrac@uaspire.org
www.uaspire.org
@uAspire
Brian Kathman
CEO
Signal Vine, LLC
(703) 338-1046
brian@signalvine.com
www.signalvine.com
@signalvine
Jessica Kennedy
Director of Communications and Outreach
WV Higher Education Policy Commission
(304) 558-0655
jkennedy@hepc.wvnet.edu
www.wvhepc.org
@jackennedy @cfwv @wvhepc