Bernie Hogan, Oxford Internet Institute & Kristen Berg, Netlab, University of Toronto. http://www.oii.ox.ac.uk/people/faculty.cfm?id=140 Connected Practice Symposium: Human Services in the Network Society, Changes, Challenges & Opportunities. The Institute for Advanced Studies, Glasgow, 14 September 2009. http://connectedpractice.iriss.org.uk/

1. Approaches to Network Collection
for Internet and Health Services
Kristen Berg Bernie Hogan
NetLab, Oxford Internet Institute
University of Toronto University of Oxford
Human Services in the Network Society
September 14, 2009
Friday, October 23, 2009 1

2. Table of Contents
• Social Networks and Health (K+B)
• Collecting Social Network Data (B)
• Considering Health and Wellness
Information (K)
• Relating Health/Wellness to Networks (K
+B)
• Conclusions (B then K)
Friday, October 23, 2009 2

3. Networks relate to health outcomes such
as obesity, STIs, and even life expectancy
Source: Christakis & Fowler. N Engl J Med 2007;357:370-9.
Friday, October 23, 2009 3

4. Health Management in Context
Network
Characteristics
Social support
+ Information
acquisition seeking
(resources)
Friday, October 23, 2009 4

5. There are many types of networks
Friday, October 23, 2009 5

6. Sociocognitive
(e.g. Recall by closeness)
Friday, October 23, 2009 6

7. Sociocognitive
(e.g. Recall by closeness)
Summer Camp
Publicly Articulated
(e.g. Facebook,
Undergraduate Family
Bebo, Twitter)
Grad school
High school
Professional
Single event friends
Professional !UK"
Friday, October 23, 2009 6

8. Sociocognitive
(e.g. Recall by closeness)
Summer Camp
Publicly Articulated
(e.g. Facebook,
Undergraduate Family
B 4 C
Bebo, Twitter)
Grad school
A
3
5
3
6
Trace data
1 (e.g. email,
2 DL
Egoschool 4
High
bulletin
1
Professional boards)
12 8
2
Single event friends
D F
E
Professional !UK"
Raw Email Network {Ego, DL, A, B, C, D, E, F}
Ego’s Neighbourhood {Ego, DL, A, B, D, E}
Ego’s Neighbourhood trimmed to symmetric
ties with in + out > 4 messages {Ego, A, B}
Friday, October 23, 2009 6

9. Networks
can show
prominence
Generally referred to in
centrality. There are
many types:
Degree - links in & out
Betweenness - shortest paths
Power - high degree friends
Closeness - easily reachable
Friday, October 23, 2009 7

10. Networks
can show
prominence
Generally referred to in
centrality. There are
many types:
Degree - links in & out
Betweenness - shortest paths
Power - high degree friends
Closeness - easily reachable
Source: Carter Butts
Friday, October 23, 2009 7

11. Whole networks cluster
Source: Jim Moody. Add Health data
Friday, October 23, 2009 8

12. Whole networks cluster
Source: Jim Moody. Add Health data
Friday, October 23, 2009 8

13. Different networks
have different
consequences
Friday, October 23, 2009 9

14. Issues with Networks I:
Standard social network analysis is done on nodes
of like type. However, information is flows through
multiple kinds of actors, all of whom are relevant
to the individual
Friday, October 23, 2009 10

15. Issues with Networks II:
Rest
Aspirin Oxycontin
Even within the standard
network paradigm, we
frequently assume “the
strength of weak ties”
and the “value of
? variety”, where accessing
Weed! Paracetemol
more information is
useful.
Job hunting (the basis of
SOWT) is different from
health info.
Friday, October 23, 2009 11

16. Issues with Networks III:
Individuals coping with sensitive issues such stigmatized
and uncommon illnesses may not find adequate support
from otherwise dense, supportive networks.
This depends on the nature of the network as much as the
structure.
Here's to Blue I'm down
What's with
Blue! Rocks with Blue
that guy? Forget it, I'm
going online
Friday, October 23, 2009 12

17. What kinds of
networks facilitate
information seeking
for Health and Medical Issues?
Friday, October 23, 2009 13

18. Information seeking
• Information flows through and is
contextualized by networks.
• However, Internet-based information
is appealing because of availability,
interactivity, anonymity and potential
to transform relationships.
• Information influences relationships
with self, health, medical
professionals, network members.
• Not all information flows in the same
way or to the same people.
Friday, October 23, 2009 14

19. Differentiating Health and Medical Information
Health information Medical information
• Information about the body and/or • Relegated to medical books and
the self supplied by professionals
• Historically devalued by medical • Historically privileged as source of
expertise well-being.
• Includes: lifestyle, wellness, and self • Includes: stigmatized concerns,
care mental health, chronic issues
• Issues that require “care” • Issues that requires “treatment”
• Dense networks facilitate common • Dense Networks of similar type
health behaviour (good and bad) support, of different type
• Diverse networks allow adoption of misunderstand
new health practices • Diverse networks allow local
disclosure
Friday, October 23, 2009 15

20. Data: Connected Lives
• Multi-stage research project in
Toronto (and Northern Ontario)
investigating ICTs and sociability.
• Data comes from East York
(Toronto) surveys and interviews.
• 350 households survey. Broadly
comparable to census tract data.
• 25% convenience sub-sample
(n=86) for interviews.
• Slightly older and less likely to be
single than in the population at
large.
16
Friday, October 23, 2009 16

21. Capturing Health Management
• Hours of Internet use
(general, health-specific)
from survey
• Health Assessment
from interview
• Network Data
discussed above
• Support data
via the network
17
Friday, October 23, 2009 17

22. Results
“How would you classify your most recent health issue?”
Lifestyle: n=59 Chronic: n=11
Health score = 7.9 Health score = 6.6
Weight loss and exercise Arthritis, diabetes, hepatitis C
0.7 hours (home) and 0.4 hours (work) 3.7 hours (home) and 0.5 hours (work)
online for information online for information
16 vc ties; 25 sc ties 18 vc ties; 22 sc ties
Acute: n=3 Mental Health: n=7
Health score = 9.3 Health score = 8
Heart surgery, cancer, dental issues Depression, attention deficit disorder,
0 hours (home) and 0 hours (work) anxiety
online for information 0.8 hours (home) and 0 hours (work)
5 vc ties; 11 sc ties online for information
11 vc ties; 10 sc ties
Friday, October 23, 2009 18

23. Searching for health information
“Iʼm sure I Googled it, cause thatʼs become a verb now. I
usually type in Canada; I wouldʼve put nutrition. … I
remember searching for nutritionist information once
and finding Ontario Canadian Website – I think
Ontario Nutritionists…. (880)”
“you know probably Google or Merck Manual or
something like that. You know if I have a medical
condition that I know about and I just want more
information about it, I would look online… (432)”
“Medical things I tend to look up myself, you donʼt want
your friends to think you are a hypochondriac (366)”
Friday, October 23, 2009 19

24. Classification of
Health Management Measures
Sick Savvy Solo
• Health score • Computer use • Network
• Medical issue • Attitude characteristics
20
Friday, October 23, 2009 20

25. What happens if youʼre all
three (Sick, Savvy AND Solo)?
• Mr. Diabetes • Mr. Frostbite • Ms. Hep C
(343) (561) (608)
Friday, October 23, 2009 21

26. What happens if youʼre all
three (Sick, Savvy AND Solo)?
• Mr. Diabetes • Mr. Frostbite • Ms. Hep C
(343) (561) (608)
Friday, October 23, 2009 21

27. What happens if youʼre all
three (Sick, Savvy AND Solo)?
• Mr. Diabetes • Mr. Frostbite • Ms. Hep C
(343) (561) (608)
Friday, October 23, 2009 21

28. What happens if youʼre all
three (Sick, Savvy AND Solo)?
• Mr. Diabetes • Mr. Frostbite • Ms. Hep C
(343) (561) (608)
Fam
Fam
Fam
Fam
Fam
Friday, October 23, 2009 21

29. Mr. Diabetes (343)
• He is a 34 year old married civil engineer with 2
young children. Has been online since 1994 and
uses his home based Internet computer about
weekly. He searches for health information
online and connects with others who have
diabetes.
• “Because Iʼm diabetic, so Iʼve visited seven or
eight times Google. Then I write down the
diabetic treatment.
Friday, October 23, 2009 22

30. Mr. Frostbite (561)
• He is a 42 year old married man with no children
who moved to Canada within a year of the
interview. He uses his Internet connected home
computer about daily.
• “I was just clicking on the condition, a search,
and then narrowing the search, until I found a
forum. Forums with people, because you find a
tht a professional or a doctor, his terminology
arenʼt the type of terms I might use”
Friday, October 23, 2009 23

31. Ms. Hep C (608)
• She is a 58 year old woman, married no children. Lived in
Canada 8 years, moved from US for better medical
system. Uses the Internet about daily and has been
online since 2002.
• Her management of a chronic health issues involved
doing research online and contributing to an international
database. She wants to know as much as possible and
has professional mechanisms for trusting the sources and
treatment options.
• “I would go to any FDA site where theyʼre doing research
and start in clinical trials and then from there you get your
key words again”
Friday, October 23, 2009 24

32. “Who in your network do you go to
for health information?”
10% of those interviewed do not mention a
network member that can provide information
42% of that 10% mention an external source,
either their doctor, or the Internet
25% of total sample mention someone outside of
their network, either a doctor or the Internet
Friday, October 23, 2009 25

33. Summarizing the analysis
• Information flows through networks and media in different
ways. Medical and health information are sought out in
different ways.
• People can add to their ʻinformation gathering networkʼ
from those outside the personal network when necessary.
• The Internet itself is considered a unique social tie that
brings with it support and information. This is especially
the case for medical information.
• This means we should expect different kinds of social
media to facilitate different kinds of information flows, and
one network (or online network site) will not fit all.
26
Friday, October 23, 2009 26

34. Summarizing network techniques
• What we learn about social structure will
depend on the technique and the type of
relationships analyzed.
–Some network types:
• Whole, partial,ego-centered
–Some relation types:
• cognitive, trace, articulated
• Networks can show clusters of activity,
prominent actors and flows of information
27
Friday, October 23, 2009 27

35. Final thoughts
• Different network structures yield different
outcomes and different needs.
• In health, these needs can refer to lifestyle/
wellness and medical information
• Some people use media as a compliment to
their personal network, others use it as an
alternative.
• We should match the network to the need,
rather than assume we need a network.
28
Friday, October 23, 2009 28

36. Thank You
Kristen Berg
NetLab, University of Toronto
;)
kristen.berg@utoronto.ca
Bernie Hogan
Research Fellow, OII
bernie.hogan@oii.ox.ac.uk
Friday, October 23, 2009 29