At CASRO digital in San Antonio, Texas: Five minutes each from four different presenter. See how the US census uses social media, and how other people use it for customer experience and market research.
Cybersecurity Awareness Training Presentation v2024.03
Social Media Data from Research to Customers - Casro Digital in San Antonio, 2014
1. RTI International
RTI International is a trade name of Research Triangle Institute. www.rti.org
Social Media for Research:
Top 5 Questions and Answers (in 5 minutes)
Joe Murphy
jmurphy@rti.org
@joejohnmurphy
@surveypost
2. RTI International
1. Can social media replace surveys?
! Short answer: no
! Longer answer: yes, if you accept that social media is like
a giant, poorly designed opt-in survey
– No limit on responses or types
– Everyone can see each others’ answers
– What’s the question?
4. RTI International
3. Does sentiment analysis really work?
! It depends!
0%
25%
50%
75%
100%
March
April
May
June
July
August
September
October
November
December
January
February
March
April
May
June
July
August
September
October
November
December
January
February
March
2009 2010 2011
Green = Kaiser Family Foundation survey:
family better off with health care reform?
Orange = Positive sentiment about
health care reform on Twitter ACApassage
5. RTI International
4. Hasn’t there been a lot of research showing social
media is really effective in research?
! Yes, but be a smart consumer. Watch out for:
– File cabinet effect
– Curse of big data
– Transparency in methods
– Face validity
"Public health researchers and hospitals
can use Facebook “likes" as a proxy for
hospital quality and patient satisfaction."
Really???
6. RTI International
5. With all these caveats, is there a point in looking at
social media?
! YES!! When it is fit for use and done ethically.
! Designing new questions and needing context
– E.g. What other topics get discussed along with certain health issues?
! Cognitive pretesting
– Crowdsourcing approach to investigate single items
! Recruitment/sample building
– Nonprobability: convenience, respondent-driven (RDS)
– Probability: if target is those using social media, can be done
! Linking social media and survey responses
– E.g. ask survey respondents if they would supply Twitter handle and incorporate
Tweets in analysis
7. RTI International
With all these caveats, is there a point in looking at
social media?
! YES!! (Cont.)
! Outreach/Dissemination
– Identify existing social media groups who can endorse or allow for
encouragement to participate
– Use social media to generate interest in research findings
! Locating
– When a phone number or address cannot be found
! Diary studies
– No need to develop new software
– Let people respond using a platform they know
! Voice of the customer, CRM, online communities
jmurphy@rti.org | @joejohnmurphy | @surveypost
8. Please tweet! #CASRO @LoveStats #MRX
Social Media as a
Customer Service Tool
Annie Pettit
Chief Research Officer
Peanut Labs
annie@peanutlabs.com
@LoveStats on Twitter
www.peanutlabs.com
:)!
14. ! Goal:
On Facebook and Twitter, help
client optimize ad reach (CTRs,
Page Likes) to a target
audience
! Target Audience:
! Ages 18-54 years old
! “Low SES” is defined as
those who have not
completed high school or
have at most completed
high school (or equivalent)
! Cigarette Smokers
Target Audience & Goal
X
Easy to target a specific age range
✔
?
15. 1. Survey: Screen respondents for
target audience demographic and
behavioral characteristics; offer
broad category interests
2. Scrape: Scrape respondent’s
Facebook data
3. Analyze: Look for consistency
between stated category interests
and actual page categories
4. Follow-up: From intersecting list,
narrow category interests to
specific “top 10” pages and ask for
specific page “likes” to extend
reach
5. Create Model & Test: Run A/B test
using test ad and feed results back
into model
6. Repeat.
Process
16. Scraping: Consent
4
Ask for consent
to scrape
Facebook data
Facebook app
again asks for
consent and
makes clear what
is going to be
scraped
18. Broad Categories (“Forest”): Specific Categories (“Trees”):
TV Shows: Much more likely to like a television
show FB page
Females: Family Guy and Sponge Bob FB pages
Males: The Simpsons FB pages
Music: Much more likely to like music FB pages FB pages of Vin Diesel and Wiz Khalifa; < 35
females: Eminem, Lil Wayne, Nicki Minaj, and TI;
< 35 males: Will Smith
Food/Restaurants: More likely to like fast-food
restaurant FB pages & much more likely to share
responsibility of grocery shopping
Taco Bell, McDonalds, Coca-Cola FB pages
Magazines: More likely to like magazines within
specific genres
Females: Cosmopolitan magazine FB pages
Males: Fitness magazines FB pages
Alcohol: More likely to like alcohol FB pages Jack Daniels/Jim Beam FB pages, and less likely
to like Absolut
Females: Smirnoff
Sports: Slightly less likely to engage with sports
FB pages
Males only: FB pages of Ultimate Fighting
Championships and the NFL
Politics: Not likely to engage with political FB
pages
Males only: More likely to have served in the
military
Employment: Less likely to work in healthcare
Shopping: no significant difference in category Amazon.com and Walmart FB pages
Low SES Smoker Profile
6
When compared to high SES smokers (p=.05, n=500)
19. U.S. Census Bureau and
Social Media
Jennifer'Hunter'Childs'
Center'for'Survey'Measurement'
U.S.'Census'Bureau'