We explore the effects of providing rewards, such as money or candy, on response rates to friendsourcing requests as well as how those rewards affect perceived relationship strength.
A Market In Your Social Network: The Effect of Extrinsic Rewards on Friendsourcing and Relationships
1. A Market in Your Social Network:
The Effects of Extrinsic Rewards on
Friendsourcing and Relationships
Haiyi Zhu
haiyi@cs.umn.edu
Sauvik Das
sauvik@cmu.edu
Yiqun Cao
yiqunc@andrew.cmu.edu
Shuang Yu
shuangy@andrew.cmu.edu
Aniket Kittur
nkittur@andrew.cmu.edu
Robert Kraut
robert.kraut@cmu.edu
4. #trending
Friendsourcing is useful and free.
No surprise, then, that friendsourcing and tools to
support friendsourcing are on the rise.
Over 50% of SNS
users asked friends
questions online.
- Morris, Teevan, Panovic, 2010
Friendsourcing has
been used to create
social systems.
e.g., Bernstein et al. 2010,
Brady et al. 2013
5. But #couldbebetter
Friendsourcing
requests often fall
on deaf ears.
- Paul, Hong and Chi, 2011
Perceived social
capital costs can be
prohibitively high.
- Rzeszotarski and Morris, 2014
How can we increase response rates
to friendsourcing requests?
7. Research Questions
Does adding extrinsic rewards affect response rates
to friendsourcing requests?
Does adding extrinsic rewards change a requester’s
perceived relationship strength with friends who
respond versus those who do not?
Does the size and nature of the reward affect
response rates and perceived relationship strength?
12. Requests
Two types of tasks: survey and document revision.
Each participant posted two requests. First was
preset Second was self-defined and posted three
days afterwards.
Participants shared each task on Facebook with a
short pitch.
Also tagged six different friends on each task.
13. Pre/post lab sessions
Two lab sessions: the first before participants posted
their first request, and the second a week later.
First session: Logistics, relationship strength scale
questionnaire for each friend they would tag.
Second session: Relationship strength scale
questionnaire for each friend they had tagged. Exit
interview.
14. Measures
Responses: Responses from tagged and untagged
friends. Integer value from 0.
Relationship strength: Pre/post change in
relationship strength. Range: -100 to 100.
Interview and comment thread data to better
understand rationales for response rates and
relationship strength ratings.
24. Summary
H1 Market model: Extrinsic rewards will provide more incentive and,
thus, increase responses. Should also reduce social capital costs.
vs.
H2 Crowd-out model: Extrinsic rewards might undermine the natural
social motives to assist friends and, thus, decrease responses.
H3 Hostile economy model: Extrinsic rewards are harmful to social
relationships by introducing market-forces into the relationship.
vs.
H4 Fair pay model: Extrinsic rewards can strengthen social relationships
by enhancing fairness.
Response
Rate
Relationship
Strength
26. Descriptive Stats
60 participants posted 105 friendsourcing requests
and solicited 630 friends to respond to their requests.
Small ($1) Large ($5) Total
Monetary 17 23 40
Non-
monetary
11 19 30
Control 35 35
105
27. Response Rate of Tagged Friends
ResponseRate
0
15
30
45
60
Control Monetary Non-monetary
Small Large
11 111
28. Response Rate of Tagged Friends
ResponseRate
0
15
30
45
60
Control Monetary Non-monetary
Small Large
11 11
30
29. Response Rate of Tagged Friends
ResponseRate
0
15
30
45
60
Control Monetary Non-monetary
Small Large
11
37
3130
30. Response Rate of Tagged Friends
ResponseRate
0
15
30
45
60
Control Monetary Non-monetary
Small Large
51
47
37
3130
31. For you, not for the candy
Large rewards attracted more responses, but
responders wanted requesters to attribute their
response to the relationship, not the reward.
Direct attribution: “I’m doing this just for you!”
Indirect attribution: “Seriously? 1 dollar can hardly
compensate my apartment rent for the revision work.
just treat me well”
32. Rewards on Response
Partial support for two market model hypothesis (H1):
•small rewards yield no greater response
•large rewards yield significantly greater response
•same pattern for both monetary and non-
monetary
Summary: People attracted by larger rewards but
attribute their response to their relationship with a
requester, not the presence of the reward.
33. Change in Relationship Strength
between Requesters and Tagged Friends
-6
-4.5
-3
-1.5
0
1.5
3
4.5
6
Control Monetary Non-monetary
Small Large
34. Change in Relationship Strength
between Requesters and Tagged Friends
-6
-4.5
-3
-1.5
0
1.5
3
4.5
6
Control Monetary Non-monetary
Small Large
35. Change in Relationship Strength
between Requesters and Tagged Friends
-6
-4.5
-3
-1.5
0
1.5
3
4.5
6
Control Monetary Non-monetary
Small Large
36. Change in Relationship Strength
between Requesters and Tagged Friends
ChangeinRelationship
Strength
-8
-6
-4
-2
0
2
4
Response No Response
37. Change in Relationship Strength
between Requesters and Tagged Friends
ChangeinRelationship
Strength
-8
-6
-4
-2
0
2
4
Response No Response
Reward
38. Change in Relationship Strength
between Requesters and Tagged Friends
ChangeinRelationship
Strength
-8
-6
-4
-2
0
2
4
Response No Response
Control Reward
39. Change in Relationship Strength
between Requesters and Tagged Friends
ChangeinRelationship
Strength
-8
-6
-4
-2
0
2
4
Response No Response
Control Reward
p=0.4
p<0.01
40. Externalizing non-response
Extrinsic rewards preserves perceived relationship
strength by providing requester with an external factor
on which to attribute non-response.
I was surprised that there is no
replies. Because I thought they
might feel a little bit better
about me. Those people they
kinda of let me down by not
sending me a reply. I got
nothing. I am hurt by it…
Control condition
“None of them seriously want
the candies. They probably
just (think) this is dumb.”
“They like the big dollars.
They don’t play around with
the little dollar”
Reward conditions
41. Rewards on Relationships
Support for fair pay hypothesis (H4):
•presence of rewards increases perceived
relationship strength in presence of response.
•presence of rewards resists lowering perceived
relationship strength in absence of response.
Summary: Extrinsic rewards have a beneficial and/or
protective effect on perceived relationship strength.
43. Overview
Friendsourcing is powerful and increasingly utilized,
but often requests go unanswered.
Introducing extrinsic rewards could help, but could
hurt relationship strength between requesters and
friends, in turn.
To that end, we tested the effects of introducing
extrinsic rewards to friendsourcing requests on:
response rates and perceived relationship strength.
45. A Market in Your Social Network:
The Effects of Extrinsic Rewards on
Friendsourcing and Relationships
Sauvik Das
sauvik@cmu.edu
@scyrusk
Presenter:
Haiyi Zhu, Sauvik Das, Yiqun Cao, Shuang Yu, Aniket Kittur
and Robert Kraut
Takeaways:
1. Extrinsic rewards, if large enough, can motivate
more response to friendsourcing requests.
2. Extrinsic rewards have a beneficial or
protective effect on perceived relationship
strength between requesters and friends.
46. Limitations
• Self-selected tagged friends biases potential
responders to stronger-than-average friends.
• Could not measure perceived relationship strength
from the perspective of the responder.
• Only tested some kinds of requests. Effects could
potentially vary across different kinds of requests.
• Better personalized pitches?