Anne Sales, Faculty of Nursing, University of Alberta.

Anne Sales, Faculty of Nursing, University of Alberta

Discuss social networks and their applicability to implementation

Describe how to measure social networks

Discuss how social networks might affect uptake of implementation interventionsSmall group exercise

Describe issues in measuring and analyzing social networksWays to manage the influence of social

networks

Both sociology and anthropology have histories– somewhat different

Date back to early part of 20th century Hawthorne wiring room studies in the

Western Electric Company in Chicago– 1920s

Very difficult to do anything other than quite small scale social network studies until computing power became widely available

What structure is evident in the arrangement of squares and circles?

SOURCE: Brandes, Raab and Wagner (2001)

<http://www.inf.uni-konstanz.de/~brandes/publications/brw-envsd-01.pdf>

*Directional graph

*Source: David Knoke

History here alsoColeman and Katz (Chicago)– study of

diffusion of medical innovation from the early 1950s Explored the networks of physicians using

a new antibiotic Early, mid- and late adopters had different

positions in social networks Data have been reanalyzed and

reinterpreted (more than once)

Opinion leader interventions Typically identify opinion leader through

surveys of participants in a given group Surveys (usually) ask questions based on the

Hiss instrument, derived from the Coleman and Katz study Please name up to three individuals whom you

would go to for information about…/whose opinion you would value about…/whom you would regard as an expert in…

Boundaries of the group are rarely clearly defined

Once identified, the interventions vary Most often education as the primary vehicle

Outcomes have been mixed Cochrane review 2006

Grimshaw et al. paper in Implementation Science 2006 Attempted to identify common opinion leaders

across health /disease conditions and across professional groups

Unable to do so Effectiveness of opinion leader approaches depends

on accurate identification of opinion leaders– or does it? May depend on other factors such as what type of

group and their reliance on expertise May depend on factors related to disease or health

problem and the evidence Vast majority of opinion leader studies have been

among physicians who may be quite different from other health care providers

Theory of planned behavior Individual level theory designed to explain

and/or predict individual behavior change (or lack of change) Largely mediated through intention to change

Important set of variables in TPB relates to social and/or professional norms (Godin et al. Implementation Science 2008)

TPB and other individual level theories are silent on where and how norms are formed or how they c an be changed

There is also an issue of “perceived behavioral control”

Perceived behavioral control

Attitudes towards behavior

Subjective or social norms

Social networks

Intervention

Intention to change behavior

Behavior

Perception of intervention

Professional norms

Issues of measurementFundamentally different level from the

individual or organizationUnit of measurement is the relationship,

not the individualRelationships require a minimum of two

people Issues of analysis

Relationships have qualities Existence Strength Direction Directness Hierarchy Embeddedness Structural properties

Holes Density

Friendship Volitional Mutual but may not be reciprocal beyond a dyad Varying content

Professional May or may not be fully volitional May or may not be mutual Professional content

Advice-seeking Often not wholly volitional Usually not reciprocal Work or professional related content

Mutual aid Similar to advice seeking but may be more mutual

and reciprocal

Establishing boundaries is importantWork units have appealTypically finite, have clear demarcationCan use lists of names from work unitMay or may not be simple depending on the

organization Different approaches to obtaining network

dataGive people enumerated lists of names with

some space to addAsk people to enumerate/name based on

specific promptsAsk for limited number (usually 3 or 4)

nominations (typically used in opinion leader surveys)

Pick an intervention from this list (one you’re at least somewhat familiar with):

Education Academic detailing Opinion leader Audit with feedback Reminders

Discuss these questions:

How would you incorporate social network measurement and analysis (SNMA) into this intervention?

What would you gain by adding SNMA?

How could you design or plan an intervention to learn from SNMA?

Reporting back from small group discussions

30 seconds on the intervention Setting (s) Relevance of SNMA Effect on intervention of adding

SNMA Design or planning issues

Response bias Recall bias What aspects or types of network do

you measure?What attributes of relationships do you

want to know about?

SNA does not use traditional individual attribute analysis techniquesFundamentally different kind of data– measures

relationships, not aspects or attributes of individuals

Learning curve for software and analytic techniques

Different level as well as type of analysis Not entirely clear how best to move between

these levelsMany social network analysts believe that

analyzing at an individual level is inappropriateMulti-level techniques may be useful although it’s

not entirely clear how to assign the level of the network

Different networks may be different levels

New information that may be quite important in understanding uptake of interventions

May offer opportunities to adapt interventions based on social network findings

Much social network analysis and research has come from organizational consulting

Relatively long history of using SNA to respond to organizational problems and issuesVarying effectiveness depending on what is

done with network information Independent of other interventions,

networks themselves can be manipulated in an organizational contextThis is not newMeasuring, analyzing, and understanding is

relatively new

SOURCE: Brandes, Raab and Wagner (2001)

<http://www.inf.uni-konstanz.de/~brandes/publications/brw-envsd-01.pdf>

*Directional graph

*Source: David Knoke

Structural holesPlaces where there should be connections but

there aren’t Weak lattices

Connections exist but are weaker or more negative than they could/should be

HierarchiesSpecific design that may or may not be

intentionalHealth care is highly hierarchical

CliquesGroups that have patterns of not

communicatingWithholding information for purposes of power

Be mindful of unintended consequences The way you ask questions conditions

the responses you getThere are different kinds of networksSome may not be relevant to what you

want/need to achieve The group of people I have coffee with may not

be highly related to the group of people I would go to for help with a work related problem

There is a very large literature on networkshttp://www.insna.org/