course #11 using social media for research: recruitment...
TRANSCRIPT
Course #11
Using Social Media for Research:
Recruitment, Intervention Design,
and Ethical ConcernsCHAIRS EXPERTS
Sarah Lillie, Ph.D.
Health Decision-Making SIG
@sbmdecisions
Minneapolis VA Health Care System
Lisa Carter-Harris, Ph.D.
@drCarterHarris
Indiana University
Sherry Pagoto, Ph.D.
@DrSherryPagoto
University of Massachusetts Medical SchoolDanielle Arigo, Ph.D.
Behavioral Informatics & Technology SIG
@daniarigo
The University of Scranton
Syracuse VA Medical Center
Camille Nebeker, Ed.D.
@cnebeker
University of California, San Diego
#SBM2017#TechSIG
Today’s Agenda
• Welcome and introductions
• Hands-on examples of recruitment & intervention with social media
• Break
• Hands-on examples of ethics issues in this area
• Brief commentary
• Panel discussion & questions
#SBM2017
#TechSIG
Leveraging Social Media for
Research Recruitment
Lisa Carter-Harris, PhD, APRN, ANP-C
@drCarterHarris
Assistant Professor, Indiana University School of Nursing (Indianapolis)
Associate Investigator, Social Network Health Research Lab, IUSON
Associate Member, Cancer Prevention & Control, IU Simon Cancer Center
Affiliate Investigator, Kaiser Permanente Washington (Seattle)
Today’s Goals
Discuss social media in general as a
recruitment tool for research purposes.
Describe the process of Facebook targeted
advertisement for research recruitment and
provide an exemplar.
Describe targeting and offer tips and pearls
for real-world use of this approach.
Twitter: @drCarterHarris
What is social media?
An interactive platform for electronic
communications, used by groups of
people to create, share, and exchange
information.
This ability to identify, connect, and
potentially align individuals can
significantly amplify messages
communicated via this platform.
Pew Research, 2016
How many people?
= 1.59 billion users
= 400 million users
= 320 million users
= 161 million users
= 100 million users
Pew Research, 2016; Business Insider, 2016
Platform Mechanisms to Recruit Advantages Disadvantages
Twitter Recruitment Tweet Retweets
• Tag People & Organizations
who would find your research
relevant or interesting who
have larger follower bases
than you
• Add a relevant hashtag (i.e.,
#LCSM)
• Consider building an online
community (i.e., The Clare
Project)
o Able to loosely
target people in a
particular field or
interest area
o Ability to ‘pay’ to
’promote’ the
tweet (gets in the
feeds of users who
don’t follow you)
o Sampling bias
o Not right for all
research projects
o Very little control
over the message
once out there in
the Twitter
universe
o People who
do not fit your
criteria will
see the
message
Facebook Facebook targeted
advertisement
o Largest social
media platform
o Sampling bias
o Not right for all
research projects
Instagram Ads (owned by Facebook so ads
are linked on both platforms)
o Ability to reach a
younger age
demographic
o Sampling bias
o Not right for all
research projects
Common Terminology A homepage is the main webpage that an individual sees each time they log into
their personal Facebook account.
A newsfeed is a list of stories that are constantly updating and displays in the middle of the Facebook users’ homepage (i.e., status updates of FB friends, photos, videos, links, application activity, pages and groups followed by the user, etc…)
What is Facebook targeted
advertisement? Facebook has the ability to ’target’ the ad
◦ by key variables in a FB users profile
◦ by things they have ‘liked’ on FB
◦ by groups to which they belong
Allows the researcher to set ‘targeting’
variables about FB users the researcher
would like to see their ad
Twitter: @drCarterHarris
Examples of Targeting in Facebook
Location (both inclusion & exclusion)
◦ Everyone in this location
◦ People who live in this location
◦ People recently in this location
◦ People traveling in this location
Ad can further be narrowed to target by country,
state or province, city, and/or zip code
Zip code can be key to reaching disparate
populations
15Twitter: @drCarterHarris
Other ways or variables on which
to ‘target’…
relationship status
education level
ethnic affinity
work industryfinancial income
home ownership
home type
household composition
parents
political affiliation
key life events
organizational affiliation
Common Terminology r/t FBTA
Impression – a “view” – represents the first time the ad is served to someone in either their desktop newsfeed, mobile newsfeed, or as a right hand column ad.
Click to website – refers to a unique FB user clicking the web link embedded in the FB ad that is redirected to the website linked to the advertisement.
Audience network – a collection of mobile applications that FB partners with to offer the advertiser (i.e., researcher) the ability to extend the reach of their ad into multiple mobile applications outside of, but connected to, FB.
Twitter: @drCarterHarris
Recruitment: An exemplar
Cross-sectional descriptive design using survey methodology
Target N = 300
Inclusion Criteria:◦ Age 55-80 years
◦ Current or Former Smoker (if former, quit within past 15 years)
◦ 30 pack-year smoking history
◦ Lives in State of Indiana
Recruitment methods:◦ FB targeted
advertisement
◦ Advertisement in high-volume readership and minority newspapers
◦ In-person recruitment Indiana Black Expo
Indianapolis Senior Housing Centers
Understanding Lung Cancer Screening Behavior: Racial/Ethnic, Gender and Geographic Differences in Long-term Smokers
Twitter: @drCarterHarris
From a recruitment perspective,
FBTA was wildly successful…
Target N = 300
Surpassed 300 enrolled participants with
completed surveys in 8 days
Final N = 465 (14 day total campaign)
Twitter: @drCarterHarris
Examples Specific to Different Topics
Research Topic or Population
Caregivers of diabetic patients
Native Americans
E-cigarette users
Epilepsy
Individuals living in rural areas
Individuals recently traveling abroad
Nursing students
Ideas to Target by Likes
American Diabetes Association
Specific geographic locations
(including zip codes)
Specific brand names (i.e., Blu)
Epilepsy Foundation
Zip code (using census data to
identify rural areas)
Location (using setting ‘people
recently in this location’)
National Student Nurses Association
Twitter: @drCarterHarris
Adapting behavioral
interventions for social
mediaSherry Pagoto, PhDProfessorFounder, UMass Center for mHealth and Social MediaUniversity of Massachusetts Medical School Department of Medicine@DrSherryPagoto
Before you start…
◎ Use target platform to point of complete comfort with features
Start a Facebook page/Twitter feed for your lab
◎Join a group on the target platform ◎Get training in the platform from your librarian,
public affairs office◎Check out the Mayo Clinic
social media webinars
Why are you using social media?
◎ Primary modality?
◎ Dual modality along with another modality (visits, text, phone, etc)?
◎ Adjunctive/secondary (primary modality is something else)?
Pagoto et al (2016) Adapting behavioral interventions for social media delivery, JMIR.
What is purpose of the social media piece?
Deliver behavioral counseling instead of clinic visits?
Provide supplemental information?
Maximize attendance and retention in the intervention?
Provide a place for study participants to communicate, connect, and support each other?
Any combination of the above?
What type of social media feed?
Host generates content
Intervention feed, participants are not expected to post (public health org)
Host and participants generate content
Intervention feed and participants encouraged to post and interact (guided patient
community)
Participants generate content
No feed, just a place for participants to interact
(unguided patient community)
Which platform?
Commercial vs investigator-developed?
Which platform does target population use the most?
Which platform has functions that I need to do this intervention?
Which platform is acceptable for this purpose to the participants?
Facebook “secret” group
◎ Invitation only
◎ Only members can view group and its content
◎ Discussion strings, post links, images, videos, live videos, polls, documents
◎ Can create events
◎ Facilitate a group chat
◎ “Share” function is off
◎ All posts appear in all group members newsfeeds
Facebook page
◎ Group and content is visible to public
◎ Entry can be via moderator approval
◎Discussion strings, post links, images, videos, live
videos, polls
◎Share function on all posts
◎Can create events
◎Posts appear in member’s newsfeeds based on
their activity on the page (less activity=fewer
posts)
◎ Contamination risk
◎ Using privacy settings you can create a private group.
◎ Discussion strings, group chats, post links, pics, videos, gifs (no retweets on private)
◎ 140 character limit to tweets!
◎ Twitter can give you archive of your feed
Target population?
Do they use social media?
Consider inclusion/exclusion criteria regarding social media habits
Experience with the platform? If not, plan to train to proficiency.
What is your intervention content?
An evidence-based intervention in traditional format (e.g., print manual)?
Yet to be developed?
Content Conversion
Each platform has ever changing norms (character limits, language, conversation style, speed of responses, use of video/images, length of videos)
Traditional materials need to be converted to posts or online articles you can link to.
Videos? Images?
Adapting content
Break content down into smallest meaningful pieces
Capture all lesson objectives
Use images
Leverage existing content that’s relevant
Link to a blog where you house a library of articles
Invite conversation
Content Library
◎This is your intervention manual
◎Should always be “current”○Don’t rely on outdated content
○Anything more than a month old = ICYMI
○Anything more than a year old feels outdated on social media
◎Do plenty of developmental work to refine posts
Group Size
◎How big should an online group be???
Size of an intervention group or the size of an online community?
◎Online communities are typically not static○They grow over time, new members constantly joining
○People can usually invite their friends
The Facebook+Friends Weight Loss Study
Pilot feasibility study of a 16-week Facebook-delivered weight loss intervention (n=80) in which participants in one condition can invite friends and family who are also trying to lose weight.
We will explore how many people get invited, whether inviting close ties =better weight loss, and greater engagement
K24 HL124366
Engagement plan
◎ Frequency of posts
◎ Group moderator○ How often will they log in?
○ What is their charge?
◎ Engagement-inducing posts
Post types used in our Facebook weight loss study
1. Soliciting Thoughts/Feelings “Don’t forget to weigh in tomorrow morning! How are you feeling about the first weigh in?”
2. Soliciting Progress Report “The most effective weight loss strategy is diet tracking. How is it going so far with My Fitness Pal?? Check out this article on why diet tracking is so darn important: link”
3. Soliciting Goals/Plans/Commitment“Today’s challenge: Eliminate ONE food item from your home (and grocery list) that you habitually overeat. Which food will get the boot?”
4. Soliciting Barriers/Problems “Weekends are the toughest for staying on calorie goal! What do you think will be your biggest challenge this weekend?”
5. Weigh in
“Happy Friday! What was your weight change for the week?”
6. Information
“Got 10 minutes? Try this 10 minute TOTAL BODY strength training for beginners video.”
7. Other
Engagement by Format (n=56 participants)
Likes/post Comments/post Total/post
Soliciting Thoughts/Feelings (17%) 2.2 (1.5) 5.6 (4.6)* 7.8 (5.0)‡
Soliciting Progress Report (2%) 3.5 (3.0) 11.2 (5.3)* 14.7 (6.4) ‡
Soliciting Goals (15%) 3.0 (2.6) 5.9 (5.5)* 8.8 (7.2) ‡
Soliciting Barriers (10%) 1.7 (1.6)† 8.2 (5.4)* 9.9 (6.1) ‡
Weigh in (7%) 2.7 (1.7) 16.6 (7.3)* 19.3 (7.9) ‡
Information (46%) 2.8 (2.2) 2.5 (2.8) 5.3 (4.0)
Other (2%) 4.0 (2.4) .8 (1.2) 4.8 (3.5)
†p<.05 pairwise comparison with Information
*p<.001 pairwise comparison with Information
‡ p<.01 pairwise comparison with Information
Engagement plan
Ice breakers
Group chats
Open ended questions
Campaigns/challenges
“Microcounseling”
Weekly progress reports
Do’s
◎Preview posts to catch mistakes or poor quality not previously noticed
◎Use software to schedule posts (e.g., Buffer) to come out at regular times
◎Have a plan for engagement data to be pulled from the platform (a programmer can help with this)
Do’s
◎Focus groups, cognitive interviews to refine posts in advance
◎Seek consultation from social media marketing professionals
◎Follow social media marketing news to stay on top of changing norms (Social Media Today is a great Facebook page)
Don’t…
◎Don’t post documents that could be converted
to posts since they will get fewer click thrus
◎Don’t heavily rely on links because most people
will not click through
◎Don’t use long posts
◎Don’t use tiny fonts
◎Don’t use poor quality images
◎Don’t assume every participant will view and
read every post
Training
◎Moderator○ Must be very comfortable with platform and its
features
◎ Participants○ Orientation sessions
can be helpful
Intervention reporting
Intervention Type
Type (host-, user- or
host- and user-
generated)
Is the social network content intended to be host-generated,
user-generated, or host- and user-generated?
Primary modality Is the social network the primary intervention modality or
adjunctive?
Purpose of SNS What is the purpose of the social network?
Participants
Experience with
social media
What is the social media experience level of participants?
Current users? Non-users?
Intervention Content
Post frequency How often will posts be made by the interventionists?
Content What is the content of the posts?
Microcounseling Will interventionists be providing counseling?
Automation Will posts be automated? If so, how many?
Chats Will moderated chats be held? If so, how often?
View termination How many participants stopped viewing posts before the end
of the intervention and at what point in the intervention?
Drop out How many participants did not attend follow-up visits?
Engagement metrics
Likes/favorites How many likes did each post get? On average, what percent of posts did each participant like?
Replies/comments How many replies did each post get? On average, what
percent of posts did each participant reply to?
Original posts How many original posts did participants make? On average,
how many original posts did each participant make?
Intervention fidelity
Page membership What percent of participants actually joined the group/page/community?
Views How many views did each post get? On average, what percent
of posts did each participant view?
Moderator login
frequency
How often did moderators/counselors log in?
Moderator
engagement frequency
What percent of participant posts received no moderator
engagement (e.g., likes, replies/comments)?
Retention
Group membership
termination
How many participants exited the group before the intervention
ended?
Other MeasuresParticipant acceptability○Barriers, facilitators
○Lurking?
○Likes/dislikes for each type of post and group as a whole
Focus groups are a must
Sense of belonging, social cohesion, social support for target behavior
Thanks!Any questions?
For help with your project, contact me at:[email protected]
You can learn more by following: @DrSherryPagoto@UMassmHealthhttps://www.facebook.com/UMASSmhealth/
Ethical Considerations with
Social Media Facilitated
Research
Camille Nebeker, EdD , MS
UC San Diego
Scripps Translational Science Institute
@cnebeker
Society of Behavioral Medicine
Annual Meeting, San Diego
March 30, 2017
CAPRICollaborative to Advance Professional and Research Integrity
Disclosure
The Connected and Open Research
Ethics (CORE) initiative is building a
national resource to support the ethical
design and responsible review of
research studies that utilize emerging
technologies. The CORE is supported
by a grant from the Robert Wood
Johnson Foundation and the UC San
Diego Chancellor’s Interdisciplinary
Collaboratory Fellowship Award.
@UCSDtheCORE | thecore.ucsd.edu
The Building Research Integrity &
Capacity (BRIC) programs promote
research integrity by educating
“research” Community Health Workers
(CHWs)/Promotores de Salud about
the scientific method and human
research ethics. The BRIC programs
have received support from the NIH,
ORI and UC San Diego.
@UCSD_BRIC | bric.ucsd.edu
Social Media DefinedAny online and mobile resource that provides a
forum for generating, sharing, or discussing ideas
and content (Gelinas et al., 2016)
• Online Communities
• Social Networking
• Professional Networking
• Content development
• Location based services
Social Media Scope
• How are social media research methods different?
• What laws or regulations apply?
• How do IRBs apply the regs and ethical principles?
• What is your responsibility when designing,
conducting and reporting research using SM
methods?
• What resources are available and how you can
contribute?
Ethical and Regulatory Challenges
• Informed Consent: How can we make consent more
informed – within the social domain?
• Risks and Benefits: How should the probability and
magnitude of harm or benefits be determined?
•By-Catch: What about the rights of people who are not
research participants (e.g., friends of friends)?
•Data Management: What are best practices for storing
and sharing health data when “public” AND not
protected by HIPAA?
Social Media “Research” Methods – What’s
different?
Researcher should know if conducting research with human subjects..
Exempt?
Expedite?
Full Board?
Subject to the same regsand norms
Operational implications may differ
IRBs may not be sufficiently familiar
Tips for Researchers
https://catalyst.harvard.edu/pdf/regulatory/Social_Media_Guidance.pdf
Ethical recruitment strategies
Check TaCoS and Privacy
Check state and local laws
Describe consent and communication
Tips for IRBs
https://catalyst.harvard.edu/pdf/regulatory/Social_Media_Guidance.pdf
Develop analogiesFamiliarity with state/fed lawsCheck compliance w/TaCoSApply Belmont principles
Imagine a Human Research Protection
System Responsive to 21st Century Science
Bloss C, Nebeker C, Bietz M, Bae D, Bigby B, Devereaux M, Fowler J, Waldo A, Weibel N, Patrick K, Klemmer S, & Melichar L. Re-Imagining Human Research
Protections for 21st Century Science. J Med Internet Res. 2016; 18(12):e329. doi: 10.2196/jmir.6634.
The CORE PlatformStakeholders in the MISST
ecosystem have access
to…
Q&A ForumNetwork members post
questions & share expertise
Resource LibraryCollect and curate best
practices including IRB
approved research
protocols and informed
consent language
NetworkConnect stakeholders
including researchers,
privacy experts,
technologists,
ethicists, regulators &
participants Demo