by Ksenia Koroleva 6 th of December, 2013 THE MEDIATING AND MODERATING ROLE OF RELATIONSHIP CHARACTERISTICS IN DETERMINING INFORMATION VALUE ON SOCIAL NETWORK SITES
Jul 08, 2015
by Ksenia Koroleva
6th of December, 2013
THE MEDIATING AND MODERATING ROLE OF
RELATIONSHIP CHARACTERISTICS IN
DETERMINING INFORMATION VALUE ON
SOCIAL NETWORK SITES
2
• Facebook is the largest database of
social information. Each day:
2.7 billion likes
300 million photos
• Stream communication allows users to
get a lot of information in a short time
• Filtering algorithms ensure that users
get the right information
• Rationality: is the information we want
the information that we need?
MOTIVATION
CAUSES AND COSEQUENCES OF INFORMATION OVERLOAD:
A QUALITATIVE STUDY
advanced
active
Information Characteristics
Network Characteristics
AMOUNT
- detail [17]
- frequency [27]
NETWORK
- size [16]
- structure [6]
VALUE
- novelty [47]
- interest [24]
UNDERSTAND-
ABILITY [7]
DISTANCE [15]
RELATIONSHIP
- strength [45]
- attraction [11]
- intensity [7]
COGNITIVE [47]
AFFECTIVE [30]
CONATIVE [15]
INFORMATION OVERLOAD
PERCEPTION
STRATEGIES AND ACTIONS
HIDE [15]people/
information
passive
DRIVING CONDITIONS
cognitive HEURISTICS
[36]
DEACTIVATE account [5]
OMISSION [7]
control of SELF-
BEHAVIOR [9]
DELETE [11]people/
information
ex-anteNETWORK
CONTROL [5]
FAILED ACTION [11]
INTERVENING CONDITIONS
Impact on Newsfeed ACTIVITY [17]
Impact on Newsfeed ATTITUDE [21]
REVERSAL [4]
Influence on SOCIAL CAPITAL
[10]
Change in INFORMATION
LOAD [12]
Change in INFORMATION
QUALITY [4]
Technology [9]
CAUSAL CONDITIONS
Skills/Knowledge [6]
Time Pressure [20]
Social Pressure [11]
Social Capital [7]
Information Longing [20]
Contact Facilitation [9]
Keeping in touch [15]CONSEQUENCES
DIS-/SATISFACTION [13]
Koroleva, K., Krasnova, H. and Gu ̈nther, O. 2010. ‘Stop Spamming Me!’ – Exploring Information Overload on Facebook, in Proceedings of the
Americas Conference on Information Systems (AMCIS 2010), AIS Electronic Library, Paper 447. 3
SOCIAL CONTEXT CUES IN SNS
Referrals/Tags
Comments
‘Likes’
Time and Place
Post Type
Verbal indicator
Relational cue
4
Profile
Information
Friends & Network
Overlap
Recent activity and
interests
DETERMINANTS OF INFORMATION VALUE
Information Value
Relationship Characteristics
Similarity
Tie Strength
Information Characteristics
Social Information
Ratings
Comments
Media Type
Photos
Links
Text
5
Which information do users value on SNS?
How do relationship and information characteristics interact with each other?
SOCIAL INFORMATION
• Non-verbal cues
• Positive emotions
• Socially acceptable behavior
• Higher effort to process
• Negative emotions
• No shared context
Salancik and Pfeffer 1978
Schmitz and Fulk 1991
Schöndienst and Dang-Xuan 2012
Ratings/’Likes’ Comments
Def.: Social Information – statements and interpretations of others in the social environment
6
TIE STRENGTH
• Limited non-verbal cues
• Alternative channels
• Low cost of maintenance
• Diversity of information
• Easy transfer of information
• Tacit information
• Relevant information
• Frequent communication
Granovetter 1982
Hansen 1999, 2002
Carpenter 2003
Def.: Tie Strength – frequency and depth of interaction (Mardsen and Campbell 1984)
7
HOMOPHILY/SIMILARITY
Def.: Homophily – tendency for friendships to form between those who are alike in some
designated respect (Lazarsfeld and Merton 1954)
- Value Homophily vs. Status Homophily
8
• More trustworthy
• Less effort in processing
• Increases with tie strength
• Affective relationships
Heterophilous ties:
• Diverse
• Complementary
• Instrumental relationships
McPherson et al. 2001
Lazarsfeld and Merton 1954
Rivera et al. 2010
• Facebook application
Objective data collected automatically (multimedia type, comments, likes)
Subjective data through a survey (information value, tie strength, similarity)
• Sample
• 141 users (52% female & 48% male, age mean: 27)
• Each person evaluates up to 6 posts (5.88 on average), randomly selected from the Newsfeed
• In total, 851 observations
9
STUDY DESIGN
INFORMATION VALUE
10
Feelings, emotions evoked by the information
6 pt ordinal scale: Like very much – dislike very much
Affective Value
Evaluation of the information per se, its value
6 pt ordinal scale: Very useful – very useless
Cognitive Value
• Correlated (0,62)
• BUT! can be empirically distinguished
% of sample
dislike very much 3.64%
dislike 9.17%
slightly dislike 17.86%
slightly like 36.08%
like 23.97%
like very much 9.28%
100.00%
% of sample
very useless 24.32%
useless 21.74%
slightly useless 17.16%
slightly useful 22.21%
useful 10.46%
very useful 4.11%
100.00%
TIE STRENGTH VS. SIMILARITY
11
tie strength
% of sampleweak strong
similarity
nothing in common 7% 0% 7%
hardly anything in common 20% 2% 22%
something in common 34% 17% 51%
quite a lot in common 4% 11% 16%
very much in common 0% 4% 4%
66% 34% 100%
• Homophily is stronger in closer relationships, correlation (0,5)
• BUT! Can be empirically distinguished
MODERATION WITH TIE STRENGTH
12
Random effects Ordered Probit, N=851
* p<0.1; ** p<0.05; *** p<0.01
Affective Cognitive Affective
Value
Cognitive
Value
Affective
Value
Cognitive
Value
Likes 0.050 (0.010)***
0.050 (0.010)***
0.051
(0.010)***
0.050
(0.010)***
0.069
(0.013)***
0.059
(0.013)***
Comments -0.012 (0.008)
-0.016 (0.008)**
-0.007 (0.008)
-0.011 (0.008)
-0.015 (0.009)*
-0.020 (0.009)**
Photos
(w.r.t.status)
0.314 (0.108)***
0.374 (0.110)***
0.317
(0.109)***
0.368
(0.110)***
0.131
(0.139)
0.222
(0.144)
Links
(w.r.t.status)
-0.033 (0.086)
0.379 (0.088)***
0.022
(0.086)
0.436
(0.089)***
-0.085
(0.107)
0.392
(0.110)***
Tie Strength 0.596
(0.084)***
0.520
(0.084)***
0.335
(0.174)*
0.264
(0.179)
Likes*Tie Strength -0.055
(0.021)***
-0.025
(0.021)
Comments*Tie Strength
0.027 (0.016)*
0.029 (0.016)*
Photos*Tie Strength 0.547
(0.234)**
0.402
(0.236)*
Links*Tie Strength 0.352
(0.190)*
0.132
(0.194)
_cut1 -1.927
(0.126)***
-0.630
(0.105)*** -1.717 (0.130)***
-0.417 (0.111)***
-1.810 (0.140)***
-0.501 (0.123)***
_cut2 -1.185
(0.103)***
0.075
(0.104) -0.966 (0.109)***
0.305 (0.110)***
-1.055 (0.120)***
0.221 (0.123)*
_cut3 -0.482
(0.096)***
0.593
(0.105)*** -0.242 (0.103)**
0.836 (0.113)***
-0.329 (0.115)***
0.753 (0.126)***
_cut4 0.559 (0.096)***
1.439 (0.113)***
0.843 (0.106)***
1.705 (0.122)***
0.759 (0.118)***
1.623 (0.135)***
_cut5 1.550 (0.109)***
2.247 (0.134)***
1.879 (0.121)***
2.538 (0.144)***
1.805 (0.132)***
2.456 (0.154)***
rho 0.160
(0.035)***
0.250
(0.040)***
0.171 (0.035)***
0.248 (0.040)***
0.164 (0.035)***
0.239 (0.040)***
R2 0.04 0.05 0.9 0.10 0.10 0.11
MODERATION 1: LIKES*TIE STRENGTH
13
NB! the values of the dependent variable displayed on the plot are calculated based on the assumption of
a continuous dependent variable (not ordinal)
MODERATION 2: COMMENTS*TIE STRENGTH
14
NB! the values of the dependent variable displayed on the plot are calculated based on the assumption of
a continuous dependent variable (not ordinal)
MODERATION WITH SIMILARITY
15
Random effects Ordered Probit, N=851
* p<0.1; ** p<0.05; *** p<0.01
Affective Cognitive Affective Cognitive Affective Cognitive Likes 0.050
(0.010)***
0.050
(0.010)***
0.045
(0.010)***
0.044
(0.010)***
0.043
(0.010)***
0.043
(0.010)***
Comments -0.012 (0.008)
-0.016 (0.008)**
-0.007 (0.007)
-0.013 (0.008)
-0.007 (0.008)
-0.012 (0.008)
Photos
(w.r.t. status)
0.314
(0.108)***
0.374
(0.110)***
0.307
(0.110)***
0.335
(0.111)***
0.309
(0.110)***
0.316
(0.113)***
Links
(w.r.t. status)
-0.033
(0.086)
0.379
(0.088)***
-0.032
(0.086)
0.397
(0.088)***
-0.024
(0.086)
0.397
(0.088)***
Similarity
(centered)
0.597
(0.047)***
0.502
(0.048)***
0.441
(0.101)***
0.272
(0.107)**
Likes*Similarity 0.003
(0.011)
-0.003
(0.011)
Comments*Similarity 0.002 (0.008)
0.010 (0.008)
Photos*Similarity 0.325
(0.122)***
0.391
(0.130)***
Links*Similarity 0.235
(0.105)**
0.257
(0.110)** _cut1 -1.927
(0.126)***
-0.630
(0.105)***
-2.088
(0.130)***
-0.676
(0.104)***
-2.086
(0.129)***
-0.674
(0.104)***
_cut2
-1.185
(0.103)***
0.075
(0.104)
-1.297
(0.105)***
0.069
(0.103)
-1.293
(0.104)***
0.074
(0.102)
_cut3 -0.482
(0.096)***
0.593
(0.105)***
-0.530
(0.096)***
0.622
(0.105)***
-0.525
(0.095)***
0.626
(0.104)***
_cut4
0.559
(0.096)***
1.439
(0.113)***
0.622
(0.096)***
1.534
(0.114)***
0.628
(0.096)***
1.541
(0.114)***
_cut5
1.550
(0.109)***
2.247
(0.134)***
1.735
(0.113)***
2.418
(0.139)***
1.752
(0.113)***
2.431
(0.139)***
rho
0.160
(0.035)***
0.250
(0.040)***
0.141
(0.034)***
0.229
(0.039)***
0.135
(0.034)***
0.218
(0.039)***
R2 0.04 0.05 0.20 0.18 0.22 0.19
MODERATION 1: PHOTOS * SIMILARITY
16
NB! the values of the dependent variable displayed on the plot are calculated based on the assumption of
a continuous dependent variable (not ordinal)
MODERATION 1: LINKS * SIMILARITY
17
NB! the values of the dependent variable displayed on the plot are calculated based on the assumption of
a continuous dependent variable (not ordinal)
MEDIATION
18
Random effects Ordered Probit, N=851
* p<0.1; ** p<0.05; *** p<0.01
Affective Cognitive Affective Cognitive Affective Cognitive
Likes 0.050 (0.010)***
0.050 (0.010)***
0.051 (0.010)***
0.050 (0.010)***
0.045 (0.010)***
0.045 (0.010)***
Comments -0.012 (0.008)
-0.016 (0.008)**
-0.007 (0.008)
-0.011 (0.008)
-0.007 (0.008)
-0.012 (0.008)
Photos
(w.r.t. status)
0.314 (0.108)***
0.374 (0.110)***
0.317 (0.109)***
0.368 (0.110)***
0.308 (0.110)***
0.336 (0.111)***
Links
(w.r.t. status)
-0.033 (0.086)
0.379 (0.088)***
0.022 (0.086)
0.436 (0.089)***
-0.022 (0.087)
0.408 (0.089)***
Tie Strength (1-strong, 0-weak)
0.596 (0.084)***
0.520 (0.084)***
0.110 (0.094)
0.115 (0.096)
Similarity (centered)
0.568 (0.053)***
0.470 (0.055)***
_cut1 -1.927 (0.126)***
-0.630 (0.105)***
-1.717 (0.130)***
-0.417 (0.111)***
-2.042 (0.135)***
-0.627 (0.113)***
_cut2
-1.185
(0.103)***
0.075
(0.104)
-0.966
(0.109)***
0.305
(0.110)***
-1.251
(0.112)***
0.120
(0.111)
_cut3 -0.482 (0.096)***
0.593 (0.105)***
-0.242 (0.103)**
0.836 (0.113)***
-0.483 (0.104)***
0.673 (0.114)***
_cut4
0.559
(0.096)***
1.439
(0.113)***
0.843
(0.106)***
1.705
(0.122)***
0.671
(0.105)***
1.586
(0.123)***
_cut5
1.550 (0.109)***
2.247 (0.134)***
1.879 (0.121)***
2.538 (0.144)***
1.787 (0.122)***
2.472 (0.146)***
rho 0.160
(0.035)***
0.250
(0.040)***
0.171
(0.035)***
0.248
(0.040)***
0.144
(0.035)***
0.229
(0.039)***
R2 0.04 0.05 0.09 0.10 0.20 0.18
MEDIATION
19
Tie Strength
Similarity
Affective Value
Sobel test statistic: 0,528 (0,06)***
Similar effect with Cognitive value
Proportion of the effect that is mediated: 0,92 affective; 0,83 cognitive
0,933 (0,06)***
0,048 (0,09)
0,565 (0,05)**
SUMMARY: 5 COMMANDMENTS
20
1. Similarity: I like it because I’m like you!
2. Likes: I like it because everyone likes it
3. Comments create information overload (need more explanation!)
4. A picture is worth a thousand words
5. I don’t necessarily like links, but they can be useful, especially if I’(m) like you!
21
THANK YOU!
QUESTIONS? COMMENTS?