making sense of strangers' expertise from signals in digital artifacts
DESCRIPTION
Introduces the use of signaling theory as a decision aid in the people sensemaking process in deciding whom to contact for expertise. This work was presented at CHI 2009 in Boston, MA, USA.TRANSCRIPT
Making sense of strangersMaking sense of strangers’’ expertise from signals in expertise from signals in
digital artifactsdigital artifacts
N.SadatShami,KateEhrlich,GeriGay,JeffHancock
Proliferation of online information
“An abundance of information leads to a poverty of
attention”
- Herbert Simon
Outline of talk
Research question
Prior research
expertise search, self presentation
The use of signaling theory as a decision aid
Study design
Findings
General Research Question
‘People sensemaking’
When looking for specific expertise using a tool, how
do individuals make sense of different information
about a stranger conveyed through digital artifacts?
Context
Finding an expert to contact
Evaluating them by by viewing
online profile
Usually only after personal
networks are exhausted (Borgatti
& Cross, 2003; Cross & Sproull,
2004)
Context of study
Prior research
Expertise search
Many tools built to find experts (Terveen &
McDonald, 2005)
Focus on finding ‘best expert’
Less attention on finding people likely to respond
Prior research
Self presentation
Selective self presentation (Goffman, 1959)
Identity claims and behavioral residue (Vizier &
Gosling, 2004)
Profiles on social networking sites (Donath, 2007;
Lampe et al. 2007)
Deception can occur (Hancock et al. 2007)
Signaling theory
Interpretive framework
Theory of communication
Process of discerning and interpreting
conveyed information
Useful for decision making under
uncertainty where deception can occur
Signaling theory
Reliable signals are pieces of information that are hard to
fake (Spence, 1973; Zahavi, 1975; Zahavi & Zahavi, 1997)
Signals in digital artifacts
Assessment signals
Quality correlated with trait
Quality is ‘wasted’ in production
Conventional signals
Need not possess the trait
Social norms and mores maintain quality
Based on Donath, in press
Signals of expertise in digital artifacts
Assessment signalConventional signal
Study: Making sense of the different pieces of Information on a profile page
Enterprise expertise locator system
SmallBlue, renamed to Atlas™ (Ehrlich et al. 2007;
Lin et al., 2008)
Convenient platform for research
Description
Mines outgoing email and
instant messaging transcripts
Data aggregator
Opt in system
Participants
Email invitation
Performed at least 20 searches using SmallBlue
131 employees, 67 responded (51.15%)
Demographics
21 countries (majority US - 43.75%)
48 males, 19 females
Average tenure 10.5 years
Majority from consulting or sales (37.5%)
Find an expert in ‘AJAX’
On a committee evaluating a new project
proposal. Need a second opinion on whether
AJAX is appropriate for the project.
Why ‘AJAX’?
Among top searches in SmallBlue
User Task
“AJAX” scenario: User Task
Shami, Ehrlich, Millen, CHI 2007, Pick me! Link selection in expertise search
Mailing listmembership
Socialbookmarkingtags
Recommendedand alternate connection paths
Corporatedirectoryinformation
Blog posts
Forum posts
Corporatedirectory selfreportedexpertise
Social bookmarks
Self reported rating data
Outcome variable (1-9 scale)
Likelihood of contacting a person
Predictor variables (1-9 scale)
Social software (tags + blogs + forums)
Social connection information
Mailing list membership
Corporate directory info
Self-described expertise
Control variables
Familiarity with AJAX
Profile data
Outcome variable
Likelihood of contacting a person (1-9 scale)
Predictor variables
Participation in social software i.e. count of: tags +
blog posts + forum posts (0-1100)
Social closeness (0-6)
Mailing list membership (0-13)
Control variables
Familiarity with AJAX
Results from rating data: Social software
For each point increase in perceived helpfulness of social
software, likelihood of contact increased by 0.33 points (p < 0.01).
“People who use dogear or IBM Forums are more likely
to reach out to the community with their questions and
their expertise and therefore I would think they would
be more likely to assist in sharing their own expertise.”
Results from rating data: social closeness
“I know the people that the system recommended to go
through. If I contact them, I'll be able to get straight to him.”
“...it wouldn't be too much of a cold call to say ‘hi, I understand
you know my colleague so and so, I'm calling you about this
other topic.’ I guess it would make me feel more comfortable
knowing that I could sort of name drop.”
For each point increase in perceived helpfulness of social
connection information, likelihood of contact increased by 0.37
points (p < 0.01).
Results from profile data (counts)
Posting one more tag, blog, or forum post increased
likelihood of contact by 0.01 points (p < 0.001).
Each degree increase in social closeness corresponds to a
0.29 point increase in likelihood of contact (p < 0.01)
Signaling theory as a decision aid
Signaling theory in ‘people sensemaking’
Focus on information that is hard to fake
More credible, reliable, less open to deception
Social software related to approachability
Social connection information related to accessibility and
verifying expertise.
Implications for design
‘Page rank’ for experts
Analyze structural patterns to find ‘answer people’ (Wesler
et al., 2007)
No systematic analysis of social software
Find others likely to respond