Call Girls In South Ex. Delhi O9654467111 Women Seeking Men
Ā
Dissertation Social Network Sites
1. Dissertation Defense
by Ksenia Koroleva
23rd of August, 2012
THE ROLE OF SOCIAL NETWORK SITES
IN CREATING INFORMATION VALUE
AND SOCIAL CAPITAL
2. STRUCTURE OF THE DISSERTATION
D. The process of social capital formation
B. Information
Characteristics and
Information Value
C. Network Construction
and Network Structure
A. Introduction: SNS as communication medium
4. SOCIAL NETWORK SITES
ā¦
are web-based services
that allow individuals to:
1) construct a profile
2) articulate a list of other users with whom they
share a connection
3) view and traverse their list of connections
and those made by others within the system.
Boyd and Ellisson, 2007
5. TIE TYPOLOGY ON SNS
ā¢ Public and private
ā¢ Synchronous and
asynchronous
ā¢ Between
unconnected users
ā¢ Without interaction
ā¢ Strength
ā¢ Roles
ā¢ Affection
ā¢ Shared physical space
ā¢ Shared social space
ā¢ Shared interests
Proximities Relations
Interactions
Information
Flows
Borgatti et al., 2009
6. THE GENERIC MODEL
SOCIAL CONTEXT CUES
USER EXPERIENCE
CONTEXT
NETWORK STRUCTURE
FEEDBACK
BENEFITS
INFORMATION
VALUE
SOCIAL
CAPITAL
Network Overlap
Breadth
Depth
Comments
Ratings
Tie Strength
Frequency of Use Type of Use
Context
PLATFORM
FUNCTIONALITY
7. NETWORK STRUCTURE
Borgatti et al. 2009
Node Level
Dyad Level
ā¢ e.g. Tie Strength
Network Level
ā¢ e.g. Network Overlap
8. EVALUATION OF COMMUNICATION MEDIA
Social
Context
Cues
Social
Presence
Theory
Media
Richness
Theory
Daft and Lengel, 1986
Short et al. 1976
Sproull and Kiesler, 1986
Diversity of
contextual
cues
Immediacy
of feedback
Type of
information
transmitted
Possibility of
Relationship
Development
9. SOCIAL CONTEXT CUES IN SNS
ā¢ Verbal indicators
ā¢ Type of information
ā¢ Referrals and tags
ā¢ Time and place
ā¢ Relational cues
Context
ā¢ Ratings or ālikesā
ā¢ Comments
Feedback
Referrals/Tags
Comments
āLikesā
Time and Place
Post Type
Verbal indicator
Relational cue
11. STRUCTURE OF THE DISSERTATION
D. The process of social capital formation
B. Information
Characteristics and
Information Value
C: Network Construction
and Network Structure
A. Introduction
āI like it because
Iā(m) like youā ā
Measuring User
Attitudes
towards
Information on
Facebook.
ICIS 2011
Proceedings
Why
information
from lovers is
more valuable
than from
close friends
on social
network
sites? ICIS
2012
Proceedings
12. Which factors
induce users to
focus on content on
SNS?
What role does
experience of
users with the
medium play?
SNS RESEARCH FRAMEWORK
Kane et al. 2012, unpublished manuscript
What is the role of
social information
when processing
information?
What is the impact of
the underlying tie
strength between
users?
Network
Information
User Platform
Social
Environment
What impact does
network overlap
have on
information value?
How does social
information differ
for different levels
of tie strength?
Source of
social capital
Locus of
Agency
13. METHODOLOGIES
Data Analysis
Data Collection
Data Type
Methodology Qualitative
Studies
Subjective
Interviews
Grounded
Theory
Empirical
Validation
Subjective
Surveys
Structural
Equation
Modeling
Regression
Analysis
Objective
Facebook
Applications
14. SYSTEMATIC VS. HEURISTIC PROCESSING
ļµ Internal cues: content
ļµ Bottom-up
ļµ Cognitive resources to
evaluate the content
ļµ High motivation
Systematic
Processing
Heuristic
Processing
ļµ External cues: heuristics
ļµ Top-down
ļµ Cognitive resources to
extract heuristics from
memory
ļµ Low motivation
Bohner et al. 1995
Ajzen 2005
Model B1
Model C2
15. MODEL B1
Experience
Contextual Information
Interactions
Social Information
Network Structure
Information Value
COGNITIVE
VALUE
Comments
Ratings
Tie Strength
Frequency of Use
Ratings * Tie Strength
Comments * Tie Strength
Post Type
AFFECTIVE
VALUE
Duration of Use
16. TIE STRENGTH
ā¢ Limited non-verbal cues
ā¢ Alternative channels
ā¢ Low cost of maintenance
ā¢ Easy transfer of information
ā¢ Tacit information
ā¢ Relevant information
ā
Granovetter 1982
Hansen 1999, 2002
Carpenter 2003
Def.: Tie Strength ā frequency and depth of interaction (Mardsen and Campbell 1984)
17. SOCIAL INFORMATION
ā¢ Non-verbal cues
ā¢ Positive emotions
ā¢ Socially acceptable behavior
ā¢ Information overload
ā¢ Negative emotions
ā¢ No shared context
Salancik and Pfeffer 1978
Schmitz and Fulk 1991
Schƶndienst and Dang-Xuan 2012
Ratings (+) Comments (-)
Def.: Social Information ā statements and interpretations of others in the social environment
18. SOCIAL INFORMATION VS. TIE STRENGTH
Tie
Strength
Social
Information
Additivity VS
- Processing effort
- Shared meaning
Sufficiency
ā
19. Ordered Probit, random effects
Robustness Check
(OLS)
ESTIMATION RESULTS
Tie Strength (TS)
Ratings
Comments
Post Type
Frequency Duration
*TS
*TS
0.39***
0.09***
-0.02
0.313***
-0.02*
0.002
Cognitive Value (1)
0.30***
0.1***
-0.06***
0.155***
-0.02**
0.01*
Value (2)
0.39***
0.11***
-0.04**
0.27***
0.21** 0.15**
-0.02*
0.01
ij
*
y =a + 1
b ij
TieStrength + 2
b ij
Ratings + 3
b ( ij
Ratings ij
TieStrength )+ 4
b ij
Comments + 5
b ( ij
Comments ij
TieStrength )+ 6
b ij
PostType + i
x + ij
e
Sample: 810 observations from 135 users;
***, **, * indicate that the estimated coefficient is statistically different from 0 at the 1%, 5% and 10% level, respectively
20. SOCIAL INFORMATION AND TIE STRENGTH
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
0 1 2 3 4 5 6 7 8 9 10
Estimated
composite
coefficient,
Ī²
s
Number of Ratings
s = 4 (very well)
3 (quite well)
2 (slightly)
1 (hardly)
0 (donāt know)
The composite impact of Ratings over different levels of tie strength:
1
b ij
TieStrength +
2
b ij
Ratings +
3
b (
ij
Ratings ij
TieStrength )
21. MODEL C2
CONTROLS
Frequency of Use
Social Information
Network Structure
Comments
Ratings
Tie Strength (current)
Posting Frequency
ATTENTION to
Information
Tie Strength (future)
Network Overlap
22. COHESION VS. STRUCTURAL HOLES
ā¢ Trust through reputation
ā¢ Security due to social norms
ā¢ Information diversity
ā¢ Access to new contacts
Coleman 1988, 1990
Burt 1992, 1997
ā
Network Cohesion
Structural Holes
23. TIE STRENGTH VS. NETWORK OVERLAP
ā¢ Corr (Tie Strength, Overlap) = 0.21
ā¢ Tie strength ā Network overlap 13%
e.g.
classmate
6%
e.g. close
friend
68%
e.g. recent
acquaintance
13%
e.g. lover
Tie Strength
Network
Overlap
High
>12%
high (1)
low (0)
All calculations are based on the data from 121 users (3025 observations)
Network
Overlap
(-)
Tie Strength
(+)
24. Partial Models
RESULTS
Tie Strength (current)
Tie Strength (future)
Network Overlap
Ratings
Comments
^2
1.02***
0.58***
-0.05***
0.41***
0.001*
0.91***
0.54***
0.46***
-0.029
0.35***
0.001
Posting Frequency
0.16 0.17
0.03*** 0.03*** 0.03***
0.14
=
ij
*
y =a + 1
g 1ij
TieStrength + 2
g 2ij
TieStrength + 3
g ij
NetworkOverlap + 4
g ij
2
NetworkOverlap + 1
b ij
Ratings ij
+ 2
b Comments +l i
PostFrequency + ij
e
Random Effects Logistic Regression, sample: 3025 observations from 121 users;
***, **, * indicate that the estimated coefficient is statistically different from 0 at the 1%, 5% and 10% level, respectively
26. Tie strength and
social information
help users evaluate
information
Experience is
associated with
information value.
Causality?
MAIN FINDINGS
Ratings
increase, whereas
comments
decrease the value
of information
Users prefer
information from their
strong ties
Network
Information
User Platform
Social
Environment
BUT with less
overlapping
networks
Social information
is more
important, the
weaker is the tie
27. PEER-REVIEWED PUBLICATIONS IN THE
DISSERTATION
B0. Koroleva, K., Krasnova, H. and GuĢnther, O. 2010. āStop Spamming Me!ā ā Exploring Information Overload on Facebook
Proceedings of the Americas Conference on Information Systems (AMCIS 2010), AIS Electronic Library, Paper 447.
B1. Koroleva, K., Stimac, V., Krasnova, H. and Kunze, D. 2011. āI like it because Iā(m) like youā ā Measuring User Attitudes towards
Information on Facebook. Proceedings of the International Conference on Information Systems (ICIS 2011), AIS Electronic
Library, Paper 26.
B2. Koroleva, K., Krasnova, H. and GuĢnther, O. 2011. Cognition or Affect? ā Exploring Information Processing on Facebook.
A.Datta, S.Shulman, B.Zheng, L.Shou-De, S.Aixin, and L.Ee-Peng, eds. Social Informatics, Lecture Notes in Computer
Science, Volume 6984, Springer Berlin/Heidelberg. 171-183.
B3. Koroleva, K. and Bolufe-RoĢhler, A. 2012. Reducing Information Overload: Design and Evaluation of Filtering & Ranking
Algorithms for Social Network Sites. Proceedings of the European Conference on Information Systems (ECIS 2012), paper 12.
C1. Krasnova, H., Koroleva, K., and Veltri, N. 2010. Investigation of the Network Construction Behavior on Social Networking
Sites. Proceedings of the International Conference on Information Systems (ICIS 2010), AIS Electronic Library, Paper 182.
C2. Koroleva, K. and Stimac, V. 2012. Why information from lovers is more valuable than from close friends on social network
sites? To appear in ICIS 2012, Orlando FL.
D1. Koroleva, K., Krasnova, H., Veltri, N. and GuĢnther, O. 2011. Itās all about networking! Empirical Investigation of Social Capital
Formation on Social Network Sites. Proceedings of the International Conference on Information Systems (ICIS 2011), AIS
Electronic Library, Paper 24. Best Paper Award.
28. OTHER PUBLICATIONS
Koroleva, K., Brecht, F., GoĢbel, L. and Malinova, M. 2011. āGeneration Facebookā ā A Cognitive Calculus Model
of Teenage User Behavior on Social Network Sites. Proceedings of the Americas Conference on Information
Systems (AMCIS 2011), AIS Electronic Library, paper 392. Best Paper Award.
Brecht, F., Koroleva, K. and GuĢnther, O. 2011. Increasing the Global Reach: Using Social Network Sites for
Employer Branding. Wirtschaftsinformatik Proceedings, AIS Electronic Library, Paper 17.
Koroleva, K., Stimac, V. and Kane, G.C. 2012. The Role of Tie Strength in Determining the Value of Information
on Social Network Sites. Submitted to Management Science.
Koroleva, K. and GoĢbel, L. Generation Facebook (working title). In preparation for JAIS.
Kane, C.G and Koroleva, K. Symbolic Action in Social Media. In preparation for Organisational Science.
Brecht, F., GuĢnther, O., GuĢth, W. and Koroleva, K. 2012. An Experimental Analysis of Bounded Rationality
Regarding risk, time and social preferences. In preparation.
In total, 7 papers in peer-reviewed conference proceedings. Part A of the dissertation presents SNS as communication medium, similar to the introduction part presented above.In part B of the dissertation we explore the impact of contextual information and social information on information value. In the paper I will present today we focus on the role of social information, and its interaction with tie strength which was presented at the ICIS2011 and nominated for best paper award.Koroleva, K., Stimac, V., Krasnova, H. and Kunze, D. 2011. āI like it because Iā(m) like youā ā Measuring User Attitudes towards Information on Facebook. ICIS Proceedings.In part C of the dissertation we explore how users construct their networks in a cost-benefit analysis and how the different measures of the resulting network structure impact the value of information users obtain through the site. In the presentation we will deal only with a latter aspect which is covered by the following paper accepted to ICIS2012:Why information from lovers is more valuable than from close friends on social network sites? In part D we explore how the characteristics of the content as well as the properties of the network interact in the process of social capital formation. The paper of this part was also presented at ICIS2012 and received a 2nd runner up best paper award.
So at first we will look into the phenomenon of social network sites: its definition, functionality, assess its role as communication medium in the widely used in IS information richness continuum. Then we will focus on presenting results from two contributions that comprised this dissertation. As the dissertation is quite comprehensive and is not possible to be presented in full, I will have to limit myself to the presentation of its several parts. In the end, we provide practical implications and discuss our findings.
SNS has penetrated our lives today: SNS has become a routine not only with younger users, but with all population groups and across boundaries. Companies pay attention to these new technologies to create value and connect with their customers. The only definition of SNS provided in literature defines SNS as web-based services that allow individuals to construct a profile, connect with other users and traverse their list of connections. *** Although the realist perspective views SNS as just another medium for people to support their relations with each other, the constructivist perspective views SNS as providing new relationships and behaviors, determined by the functionality of SNS. Specifically, friendship on SNS is a new kind of relationship which can co-exist with and impact the real-life relationship. This is illustrated by such new forms of behavior as ālikingā, or traversing the history of interactions with others in the timeline. Moreover, on SNS ties are all the same and not differentiated by their strength. Underlying real-life relationship usually determines the strength, or can be approximated by the network data. The concept of a friend thus becomes blurred as a result. ļ Therefore it becomes crucial to study these new forms of connections as well as explore, whether the rules that govern traditional real-world relationships also apply to SNS. This definition is focused on the individual nodes (users) and ties that connect them. However, there are several extensions to this definition to account for a more complex phenomenon of SNS: Nodes are not just individual users, they can represent any type of entity such as groups, organizations or information entities. Ties between nodes can encapsulate many different types of connections, such as interactions, relations and flows. Connections of a user to others are rather a backbone of communicating and sharing information.
Real-life relationships of users are usually based on physical social proximity, which is enhanced by numerous interactions, during which information flows from one user to another. Previous generations of IT-enabled platforms (e-mail, discussion board) typically supported one type of tie ā interactions. SNS, however, support a richer range of possible relational ties, specifically: Relations: ties that reflect persistent connections, which can be role-based (friends, family), affective (like, dislike), differentiated by strength (weak-strong)Interactions: discrete, transitory relational events. Relations increase the probability of interactions, interactions can increase or change relations. Flows: tangible and intangible material that can move from one node to another when nodes interact. Proximities: shared physical space (location) or social space, such as co-membership in groups, shared interestsSNS: location-based services *** SNS alters the expectations of how ties relate to one another. For example, social media facilitates relationships among distant or previously unknown people and it enables information to flow without interaction and between unconeected users. For example, on the Newsfeed information is usually shared without active interaction and flows also to the outer circles of friends. Thus, SNS represent unique and complex environments to study. *** on SNS people do not have to share a physical space, although locaiton-based services are becoming one of the main features as well.
FEW WORDSAll of the papers that comprise the dissertation can be summarized in the generic model. This generic model depicts the process of social capital formation on SNS: the functionality offered by the platform combined with the patterns of use that users develop on SNS encourage the accumulation of information and network sources, which under favorable conditions can translate into benefits, be it either valuable information or social capital. In the chapters of the dissertation, we differentiate between the main sources of social capital benefits, namely network structure and information characteristics in their impact on the benefits of social capital. Moreover, we explore the impact of user characteristics reflected in the experience with the medium as well as social information on informational and other benefits, whereas implicitly considering the underlying platform functionality. On a micro-level, the model can be used when evaluating any piece of information that is encountered on the network. As such, this information is accompanied by several informational properties, such as: the breadth, the depth, and contextual features. Additionally, the person who is sharing this information is assessed, specifically the underlying relationship and the overlap in the network. Social information from other users in the network is taken into consideration, as it provides additional cues to estimate the value of information. In the background, user characteristics, such as the experience of users with the medium or in communication with the person who is sharing the information might accelerate or constrain the assessment of information. At the same time the underlying network functionality determines breadth and depth of information, as well as allows to visualize the network.
SNS offers unique possibilities to visualize the networks of users and thus analyze them on different levels.If researchers wanted to previously assess the networks of users, they had to ask users to name several people and describe their relationship with them. Of course such an estimation could not be considered objective.Six degrees of separation refers to the idea that everyone is on average approximately six steps away, by way of introduction, from any other person on Earth, so that a chain of "a friend of a friend" statements can be made, on average, to connect any two people in six steps or fewer.Researchers in the area of Social Network Analysis (SNA) usually study the configurations of individualsā networks on three different levels: at the network level by analyzing the structure, measured by e.g. network cohesion or availability of structural holes. at the dyad level by studying the relationship between two people, strength determines their relationship; at the node level by estimating the structural position of a person, with the help of e.g. a centrality measure (Borgatti et al. 2009).The structure of the netoworklargerly determines the amount and type of benefits that can accrue to the users due to the maintenance of their relationships with others.
In order to evaluate communication media, a variety of theories is employed. Among the most known of these are Media Richness, Social presence and social context cues. These theories point to the same causes and effects regarding the impact of contextual cues and the immediacy of feedback on the type of information that is transmitted and the subsequent possibility of relationship development. Due to its ability to transfer less cues in general and less social information in particular as well as exhibit significant delays in feedback previous forms of CMC (such as E-mail) are considered less rich than FtF communication and therefore exchanges through it can only be beneficial for sharing less complex and rather objective information. The absence of cues makes media more impersonal and even depersonalizing (Walther 1992, Sproull and Kiesler, 1986) and therefore communication through it can impede relationship development (Walther 1995). Media richness ā defined ability of media to provide understanding within a time interval ā especially stresses the role of the immediacy of feedback for transferring complex information and thus evaluation of communication media. Social Presence: degree to which a medium facilitates the awareness of the other person during the interaction and the consequent salience of the interpersonal relationship ā streses multiplicity of contextual cues, such as the ways in which information can be transmitted ā text, verbal and especially non-verbal ones. Although several dimensions of social context are important, situational context cuesare most salient for subsequent relationship development (Sproull and Kiesler 1986), which include the relationship among senders and receivers, the topic of communication and the norms and conventions that define the nature of the social situation (Walther and Burgoon 1992, Sproull and Kiesler 1986). ļ We would like to analyse SNS based on these two dimensions: the ability to transmit contextual cues and provide feedback to establish its place in the media richness continuum.
CMC was criticized due to the lack of contextual cues. However, on SNS there are a lot of contextual cues, which we subdivide into: contextual information, relational and social information. Contextual information: Although in traditional CMC it was possible to attach documents to the E-mails, on SNS various types of content can be shared more effectively ā pictures, links and videos. These alternative types of information help to transfer information more effectively (compare the effort necessary to process pictures as opposed to text), as well as transmit a lot of tacit information ā information that is difficult to put in words ā critical to establish shared meaning. At the same time, verbal indicators can also convey contextual information : lexical surrogates provide for informal communication, whereas various linguistic cues can express immediacy of communication (Walther 1992). SNS users use a lot of linguistic surrogates, such as hearts and smileys that help to transfer their emotions and intentions.What is most unique in SNS are referrals can call the attention of other users to the information that is shared and thus ensure that the information reaches the people to whom it provides most value (Burt 2002). At the same time, each post on SNS is saved with time and place which provides additional cues to the recepients.Social information or feedback from others. Social information is defined as the perceptions of others in the social environment about the information that is shared (Fulk et al. 1987), It also serves as the feedback from others on the information they receive. Through feedback mechanisms provided by the platform, individuals learn from others (Bandura 1977) and adjust their behaviors accordingly (Bandura 1963). Ratings are standardized mechanisms that allow users to evaluate their perceived value of information on a standardized scale, such as one-sided affirmations (e.g. Facebook ālikeā), binary decisions (e.g. Digg up or down), or along a continuum (Amazonās 1-5 stars). On such SNS as Facebook mainly one-sided affirmations are used, as providers fear propagation of negative feedback. Comments are open-ended mechanisms that allow individuals to register their more elaborate opinions on certain digital content. Although these two types of feedback are sequential, ratings from other users are akin to non-verbal responses (such as thumbs up or down) and therefore might partially play the role of concurrent feedback. Especially the one-sided ratings (likes) signal to the receiver that there is a certain number of users who have understood and agree with the information shared. ļ Although in the following we mainly focus on social information, contextual and relational information are assessed in the paper as well. SNS, however, might challenge this proposition by providing a lot of contextual information to its users. Contextual information allows to process information more quickly and effectively (Dennis and Kinney 1998), as well as establish the shared meaning quicker (Miranda and Saunders 2003). Moreover, a lot of the information shared on SNS is ambiguous without understanding its context, where the role of contextual cues for interpreting this information becomes particularly acute.
WE PRESENT TWO PAPERS OUT OF EACH PARTThe first paper is quite a long running project, which started with my PhD and has allowed me to choose the specialization and to achieve what I have. The second one is more new, but it shows how one can combine the objective data and subjective evaluations of users. Part A of the dissertation presents SNS as communication medium, similar to the introduction part presented above.In part B of the dissertation we explore the impact of contextual information and social information on information value. In the paper I will present today we focus on the role of social information, and its interaction with tie strength which was presented at the ICIS2011 and nominated for best paper award.Koroleva, K., Stimac, V., Krasnova, H. and Kunze, D. 2011. āI like it because Iā(m) like youā ā Measuring User Attitudes towards Information on Facebook. ICIS Proceedings.In part C of the dissertation we explore how users construct their networks in a cost-benefit analysis and how the different measures of the resulting network structure impact the value of information users obtain through the site. In the presentation we will deal only with a latter aspect which is covered by the following paper accepted to ICIS2012:Why information from lovers is more valuable than from close friends on social network sites? In part D we explore how the characteristics of the content as well as the properties of the network interact in the process of social capital formation. The paper of this part was also presented at ICIS2012 and received a 2nd runner up best paper award.
Since their launch, researchers in various disciplines have been studying various questions surrounding the emerging phenomenon of SNS. In IS, a significant body of literature is dedicated towards understanding the motivations behind SNS use, the impact of intensity of SNS use on social capital,privacy concerns,impact of profile features, sentiment of interactions. However, these studies usually explore just one question connected with SNS without assigning it to the overall framework. What most of them fail is to address such questions as: Why is it important to study SNS? How do these environments differ from other forms of CMC-communication? How do they relate to the real-world behavior of users and impact their lives? Addressing these important questions, Kane et al. (2012) propose a framework which structures all research questions relating to SNS along two dimensions: source of social capital and locus of agency. The source of social capital is important in the sense that it allows to differentiate whether the benefits of a network stem from the resources that flow in the network (i.e. information) or through properties of the network structure (Nahapiet and Ghoshal 1998). The locus of agency, on the other hand, allows to differentiate whether benefits are rather determined by the users or the platform. Or the social environment that we add. GENERALWhat factors induce users to focus? What role does experience play? SPECIFIC: (Because SNS allows to so effectively visualize) What is the role of netowork properties (Because social information is one of the most important determinants of behavior) What is the role of social information?In order to structure the research questions that we address in this dissertation, we adopt and extend this framework. First of all, we explicitly differentiate between dynamics vs. outcomes (Kane et al. 2012). The former is rather concerned with how networks and information forms, whereas the latter with outcomes to which these properties lead. Moreover, considering that these environments involve not only the individual in question, but also connections to and interactions with other individuals on the platform, which impact behavior of the user in question, we add social environment to the locus of control dimension. Therefore we can position each research question we address in a three-dimensional space with the following axes: a) source of social capital (content vs. network); b) the dynamics (sources vs. benefits); and c) locus of control (user vs. platform vs. social environment). On the one hand, users may use SNS platform in very different ways than those intented to or predicted by the designers of the platform AND on the other hand, features of SNS platform may support certain digital content and structural capabilities that will determine the networks that form on those platforms. User: how individual users build, manage, and benefit from personal networks developed on particular SNS platforms and content that is created on them. Therefore, such questions as: which content the users focus on? Which is the role of tie strength in determining information value?Platform: how media platforms influence the formation and dynamics of social networks they support. By platform we mean technical systems that enable the network and sociological and economic aspects that govern the network environment. What is the general role of the experience with the medium? As network overlap is easily calculated and presented on the network, how does it influence the behavior of users on the network?Social Environment: how social environment determines the outcomes and behaviors. What is the role of social information as well as how does the impact of social information change, depending on tie strength?Dynamic theories are concerned with how networks form and acquire their particular properties Outcome theories are concerned with concequences of network properties.
Advantages and disadvantages of collecting the data through FB:(-) for some variables, we need to collect additional response (such as tie strength), as it can not be measured by the objective data (-) restricts the people who answer (those who have FB, those who are FB users, those who do not have privacy concerns) (-) the data is too raw, needs assumptions on its aggregation (+) Allows to elicit real behavior of users (as opposed to stated one), diminishes response bias (like overstating, hiding negative things)(+) Allows to measure variables that are hard to measure by asking respondents directly (how many did you unfriend)(+) Simulates the real environment, so that the users do not have to imagine āhow toā, but have answers already preparedFEW WORDSIn most of the studies we present we use methodological triangulation. First of all, we use triangulation between qualitative and quantitative methodologies. That is, we use qualitative methodologies at the exploratory stage to generate the conceptual models, and then use different quantitative methodologies to empirically test them. While the main aim of the GT approach is to build theory (e.g. expressed in a set of propositions), other methods can be used to verify it (Pace 2004). Additionally, the value of the dissertation lies that it is not only based on the stated behavior and preferences of the users, but partially also on their real behavior on the network. That is, subjective behavior is usually collected in surveys, whereas objective ā through specifically designed facebook applications. As such, we design and implement applications to provide users with a real experience and additionally collect real data about their behavior on the network. Application setting, in its turn, allows to manipulate different designs. For example, in the first application users were presented with posts one by one, and asked to evaluate them. Here they were induced to process them rather systematically. In the second application, however, users were presented with 25 posts all together and asked to choose the ones the would pay attention to, rather simulating heuristic processing. ***SEM has the advantage that it is best suited to test the exploratory models ā the ones that we generate based on the qualitative data. At the same time SEM has the advantage that it can be used with non-normally distributed variables, which is the case for many of the variables we collect in surveys. Moreover, SEM is best suited to test such latent constructs as attitudes and evaluations of users, which are mainly the focus of our analysis. *** Panel regression: accounts for the unobserved respondent-specific effect by including an additional error term into the equations.Most of the models are tested via several different methodologies to ensure that our findings are robust and the identified patterns are present in the data, rather than driven by the employed methodology. If a regression model is tested, it is always verified with an alternative: for example, if a random-effects specification is used, it is verified by a fixed-effects one. ***The use of fixed effects offers the advantage of robustness in the presence correlation between the set of explanatory variables and the respondent- specific effect. Random effects, on the other hand, is useful if the unobserved respondent-specific effect is correlated with other omitted effects that are captured by the error term. If we want to delineate the impact of two variables with an opposite impact on the dependent variable, we test the regression by excluding each of them and comparing the coefficients. If they get closer to 0, we are able to show the effect of the omitted variable bias, and thus inversely argue for the necessity to include this variable in the analysis.
Systematic processing Dependent variable: cognitive and affective value of information, model B1 Systematic processing is a bottom-up approach, involving extensive evaluation of arguments and issues involved in a message (i.e. its content) and comparing that information to existing knowledge structures and beliefs (Bohner et al. 1995) in order to arrive at an evaluative judgement. For systematic processing of information, a significant amount of motivation, ability and cognitive resources are requiredHeuristic processing involves just screening the newsfeed for the content to pay attention to. Dependent variable: attention to information, explored in the model C2In contrast, the top-down heuristic processing strategy involves reliance on certain cognitive heuristics ā rules of thumb, schemas or other stereotypes ā to evaluate information. Cognitive heuristics are mental shortcuts that allow people to form opinions without extensively analyzing the contents of the message based on certain cues present in the situation. Cognitive heuristics are gained through past experiences and observations, stored in memory and activated when the message reflects a certain feature ā a heuristic cue - that signals its relevance (Chaiken 1980). Thus, the individuals form their opinions quickly and efficiently without engaging in extensive evaluation of the content. ***Examples of widely employed heuristics, confirmed in numerous experiments, include: āconsensus implies correctnessā (Maheswaran and Chaiken 1991), āpeople agree with those they likeā (Chaiken 1980), ālength implies strengthā (Wood et al. 1985), or āexpert statements can be trustedā (Chaiken 1980).
HOW WAS MEASURED ā APPLICATION 1 DEPENDENT VARIABLEINDEPENDENT VARIABLES The hypotheses you have in the handout. In the model B1 we explore the impact of social information and tie strength and their interactions on the affective and cognitive value of information. This model is based on the first application, which upon permission of the user, took the information directly from the userās newsfeed and presented 6 posts to the users one by one and asked them to evaluate these on the affective and cognitive dimensions. Additionally, users were asked about the strength of the relationship with the person whose post they were evaluating. As people were processing information rather systematically, we assess the different dimensions of the subjective value of information. Value beliefs are complex and can possess multiple dimensions, such as: intensity, importance, knowledge, accessibility and affective-cognitive consistency (Crites et al. 1994, Voss et al. 2003). Generally accepting the critique of the uni-dimensional structure of beliefs (Voss et al. 2003) most authors differentiate between affective and cognitive components of value (Ajzen 2005; Voss et al. 2003; Yang and Yoo 2004). Cognitive value refers to evaluations of the qualities of the information itself, whereas affective value focuses on how much the person likes the information and is emotionally attached to it (Ajzen 2005). B1.1: the strength of the tie between the respondent and the source of information relates positively to the perceived affective and cognitive value of information provided by that source.B1.2: ratings relate positively to the perceived affective and cognitive value of information on SNS.B1.3: comments relate negatively to the perceived value of information on SNS.B1.4: the weaker the relationship with the source of information, the more influential are the ratings and comments provided by other users to evaluate information from this source.B1.5: Pictures will have more affective and cognitive value than other types of content, such as text, on SNS.B1.6: the frequency and duration of SNS use will relate positively to the value of information.Post type as a form of contextual information: we assume that pictures will be more easily processed and preferred on the network, as opposed to other types of content. Frequency and duration of use, although they characterize only the respondent (and we do not include them into the basic panel regressions, as the additional error term accounts for the variance in each respondentās characteristics), will be related positively to information benefits.
Many studies have investigated the influence of tie strengthādefined as the frequency and depth of interaction (Mardsen and Campbell 1984)āon the individual ability to obtain valuable information from a social network. more likely to find valuable information about job searches through weak ties than strong ties (Granovetter)novel information and more diversified information. Burt 1992The advent of IT-enabled communication networks (E-mail) lead the researchers to believe that weak ties have more value in these networks (Constant et al. 1996, Pickering and King 1995). Researchers initially considered weak ties because: they are primarily capable of communicating relatively thin types of content, such as text (Daft; Pickering, Orlikowski)and do not transmit a lot of contextual cues. Alternative channels for communication with strong ties, therefore prefer SNS for weak ties Low cost of maintenance on SNS ā can communicate with people simultaneously, or just have others in the network: SNS were originally related to increased in both bridging and bonding social capital, but bonding later was disproven.SNS features that undermine this: Richer forms of communication, multimedia and hypermediaSupport existing relationships rather than new onesStrong Arguments:- First, strong ties are associated with better transfer of information (Hansen 1999). Social media platforms provide features that automate many aspects of the search process (e.g. NewsFeed, Twitter lists), lowering the value of weak ties for information search.Strong ties, on the other hand, are more willing to share information and to devote their time to assist one another thus creating a favorable environment for information transfer (Coleman 1988, Uzzi 1996). Second, strong ties are also particularly helpful for transferring tacit information, information that is difficult to put into words (Hansen 2002).Moreover, strong ties possess knowledge about who knows what and requires which information and therefore are more valuable in exchanging information (Uzzi 1997, Hansen 1999). Research suggests that in situations with abundant information, people prioritize information provided by strong ties (Carpenter 2003). NB! Channel expansion theory***But more frequent communication among parties also might enable them to develop a shared understanding of the medium and thereby functionally expand the capacity of these channels for rich communication (Carlson and Zmud 1999).***The later empirical evidence finds that both the diverse network of weak ties and a high bandwidth of communication with strong ties can provide users with diverse and non-redundant information, depending on the environment surrounding these ties (Aral and Van Alstyne 2011). Recent findings: the later empirical evidence finds that both the diverse network of weak ties and a high bandwidth of communication with strong ties can provide users with diverse and non-redundant information, depending on the environment surrounding these ties (Aral and Van Alstyne 2011). However, the relative informational value of strong versus weak ties may depend on other factors, such as the pursued goal (Hansen 1999), type of information (Uzzi 1997), task (Rowley et al. 2000), or organizational structure (Reagans and McEvily 2003; Oh et al. 2006).
Social Information: Social information stemming from the behavior, statements, interpretations and cognitive assessments by others in the social environment (Schmitz and Fulk 1991) provides cues for users to develop their attitudes, opinions and even needs (Salancik and Pfeffer, 1978). As opposed to rational choice models such as media richness, the social influence model (Schmitz and Fulk 1991) postulates that individual perceptions about media are, in part, socially constructed, as they are impacted by the different evaluations of media characteristics provided by others.Ratings:. EASY TO PROCESS.Thus, they make certain information more salient, by providing cues about which information to consider (Salancik and Pfeffer 1978). Non-verbal cues:Specifically, ratings are similar to nonverbal cues, such as a thumbs up or down, and thus provide instant impressions of the information being shared while also promoting awareness of other people who interact with the content on the platformSocially acceptable behavior: If presented with the opinions of others about something that are in consent, people agree (Chaiken et al. 1989). Herding: If everyone likes it, I like it tooā. Positive emotions expressed in the post (SchoĢndienst and Dang-Xuan 2012)Comments Comments instead are akin to verbal responses, which must be processed extensively. Although they provide more elaborate evaluations of shared information, they might cause information overload and are therefore likely to exhibit a negative association with information value. First, their verbal nature and the inability to summarize comments as effectively as ratings may create an information overload. Because SNS users often contribute comments simultaneously and are not limited in the number of verbal symbols they use, more information gets exchanged on the network (Dennis 1996; Schultz and Vandebosch 1998). More contributors tend to lead to decreasing marginal value (Asvansund et al. 2004; Schroder et al. 1967); that is, additional comments require similar amounts of information processing but produce less insight. As a form of verbal feedback, comments can clarify content, complete a statement, or express a controversial opinion (Dennis and Kinney 1998). Empirical evidence suggests that on SNSs, high numbers of comments imply negative emotions expressed in the post (SchoĢndienst and Dang-Xuan 2012). Therefore, information that sparks many comments likely signals controversy and may decrease its overall credibility. Third, users might be reluctant to rely on the opinions of others who are not similar to them (Salancik and Pfeffer 1978) or with whom they do not interact frequently (Erickson 1988). Those who comment usually represent the social network that surrounds the source of information, which increases the probability that the respondent who is evaluating the information is not part of this network or is only weakly connected to it. HOWEVER: the impact depends on how people process information. If information is processed systematically, comments will have a negative impact. If information is processed heuristically, then they might have a positive impact on information value.
Because social information cannot be separated from the source of the information (i.e., the underlying relationship between the source and the respondent), we must consider how these contextual cues interact to influence information value on SNSs. Additivity:On the one hand, tie strength and social information might have additive impacts on evaluations, such that the two cues lead to higher evaluations than if just one was considered (Chaiken 1980). In this case, the positive effect of ratings on information might be enhanced by a stronger relationship with the source of information. On the other hand, the negative effect of comments could be offset (or exacerbated), depending on the strength of the relationship. Sufficiency:If one contextual cue delivers sufficient information, other cues might not matter. Specifically, tie strength should be the primary determinant of information value, such that people prefer to focus on their stronger ties (Carpenter et al. 2003), with whom they share meaning and can evaluate information easily. However, if time and motivation remain, they may consider information from their weaker ties, and as this information demands more effort to process, they would increasingly use ratings and comments from others in the network to evaluate it. First, social information requires more effort to process, whereas a relationship is salient and prompts automatic assessments. Second, With weak ties, the lack of shared meaning requires users to rely on other available cues, such as social information. Therefore, we hypothesize: H4: The weaker the relationship with the source of information, the more influential are the ratings and comments provided by other users to evaluate information from this source.
As each user evaluated up to 6 posts, we estimated a panel ordered probitregressin, specifically tailored to use with ordinal variables. The latent variable governs the observed variables and we estimate the cut-off points. We estimate it with random effects, as we assume that random disturbance term can be correlated with the respondent-specific random term, and not with our explanatory variables. Tie Strength: However, on SNSs, users place more value on information from strong rather than weak ties. much larger networks tend to exist on SNSs, which would increase the costs of information processing from weak, rather than strong, ties. the established shared meaning with strong ties may improve peopleās ability to derive more value from these sources of information. Ratings vs. Commentsthat these types of social information have differential impacts on the value of information on SNSs. Nonverbal ratings increase the value of information; verbal comments decrease it in such environments. Cost of processing Direction of signalsInteractions hypothesis received only partial support, in the case of ratings (the interaction of comments and tie strength is only significant for cognitive value)We find that tie strength moderates the measured relationship between the number of comments and ratings and attitude. Specifically, in the case of ratings, as tie strength increases, the positive relationship with attitude diminishes, which can be inferred from the negative point estimates of ratings*tie_strength, which is significant at 10% for affective, and at 5% for cognitive valuations.Specifically, people tend to process information provided by strong ties and weak ties differently, even though the strength of these relationships is not embedded in or reported by the system. Here we observe the so-called relationship-primacy effect: if the tie is very strong, it is enough to evaluate the information highly. The weaker is the tie, the more will the users utilize ratings to evaluate information. Contextual information: post typeWe find the affective value of pictures to be higher than that of any other information types exchanged on the network (Table 13), likely because pictures are able to depict large amounts of data effectively (Bederson and Schneiderman, 2003) and thus offer more value for minimal processing costs, compared with other post types. Experience: frequency and durationPerhaps users who are more interested in the type of information provided by the platform simply use the platform more. Another explanation instead might entail an extension of channel expansion theory (Carlson and Zmud 1999): Although people do not necessarily expand the channel through their shared usage, they may spend more time cultivating networks and learning conventions for effective communication using SNS, such that they expand the channel through more effective network management. IN order to account for these specific within-person differences, we also test a OLS model with frequency and duration. However, here we might suffer from omitted variable bias.
As comments hypothesis only received partial support in the interaction hypothesis, we will depict the interaction effect of ratings. With Figure 2 we depict the total estimated composite effects of tie strength and ratings for affective value. We basically take this equation, put in the values for Betas and calculate one each for each tie strength level, varying the number of ratings.The intercept is the corresponding tie strength * coefficient on tie strength. The slope of the line represents the coefficient on ratings and the interaction effect of ratings and tie strength. When we hold everything else constant, affective value is consistently high when the posts come from the respondentās strongest ties; the number of people who rate the information does not change this status significantly. As tie strength decreases, the marginal impact of ratings increases though. At the lowest value of tie strength, the slope of the total impact is at its steepest. At the same time, tie strength plays a more important role: note the impact of 10 comments and no tie strength vs. the impact of high tie strenght and no comments. So the full equation for the graph is the following (we take it directly from our previous equation! Itās the ceteris paribus joint effect of tie strength and ratings!)Ī²1āTieStrengthij+Ī²2āRatingsij+Ī²3āRatingsijāTieStrengthijWhere ā when we take estimates for Affective ā Ī²1 equals 0.391, Ī²2 equals 0.091 and Ī²3 equals -0.019. We can therefore also write the equation as:0.391āTieStrengthij+0.091āRatingsij-0.019āRatingsijāTieStrengthijNow, what we plot in Figure 2 is simply this relationship, for varying levels of tie strength.Ā Tie strength = 00.391ā0+0.091āRatingsij-0.019āRatingsijā0=0.091āRatingsijTie strength = 10.391ā1+0.091āRatingsij-0.019āRatingsijā1=0.391+0.071āRatingsijTie strength = 20.391ā2+0.091āRatingsij-0.019āRatingsijā2=0.782+0.051āRatingsijTie strength = 30.391ā3+0.091āRatingsij-0.019āRatingsijā3=1.174+0.031āRatingsijTie strength = 40.391ā4+0.091āRatingsij-0.019āRatingsijā4=1.154+0.011āRatingsijAnd thatās how we get the five lines in Figure 2.
FB APPLICATION DEPENDENT VARIABLE INDEPENDENT VARIABLEs: subjective vs. objective measure of tie strenght. The hypotheses you have on the paper, I will go into the most interesting ones.In this model we explore the impact of different measures of network structure on the attention of users towards information on SNS. On the dyad level of analysis, we elicit the subjective evaluations of users about the existing tie strength as well as the desire to develop a relationship with the person whose information is presented. On the network level, we objectively measure the network overlap of two users which will allow us to estimate their relative network density. We control for the social information in form of ratings and comments as well as for the experience factor exemplified by the frequency of communication. The application we design and implement presents users with 25 posts from their newsfeed and asks them to choose the ones they would focus on. The set-up of the application reflects the top-down approach and induces users to evaluate information heuristically.Therefore, the dependent variable of our study is the attention of users towards the information that is shared by their friends on SNS. That is, only the information that attracts userās attention in the overall information flow is the only valuable information that the user can effectively use. OBJECTIVE VS. SUBJECTIVE MEASURE OF NETWORK STRUCTUREIn the background other objective data is collected by the application: number of ratings and comments each information received, number of mutual friends between the respondent and the source, as well as the posting frequency of each user in question. Tie Strength (future) The concept of tie strength need not be understood unilaterally. In fact, the three necessary and sufficient conditions of a relationship between two people are: i) somewhat frequent interaction; ii) usually a mutual affection; iii) a history of interaction that has lasted over an extended period of time (Krackhardt 1992). Ties are considered weak, if they lack either the history of interaction and/or the mutual affection. Interestingly, tie strength is usually rather measured by the recency of contact or frequency of communication, but rarely by its affective dimension (Krackahrdt 1992). However, affection usually determines the relationship: if there was no mutual affection, there would be no need to interact and, therefore develop a relationship. As relationships are not formed instantly, affection for the large part is a catalysator of interaction and relationship development. It determines those weak ties that can become strong in the future, given the sufficient number of exchanges, from those weak ties that will most probably remain weak forever. Feedback: The comments and ratings are shown merely as a number, but their sheer presence can be recognized while scrolling down the Newsfeed. Therefore in this condition of information overload created by the posts in this study, we hypothesize will not dwell into determining the specific impact of ratings and comments, but be simply attracted by the information that has received some feedback as opposed to the one that has none. Posting frequency:Especially the active usage of the platform determines its value: that is, if users post a lot themselves, they perceive the medium as a useful means of communication and information exchange (or vice versa). At the same time, as more information is shared on the network, the shared meaning of that information develops (Miranda and Saunders 2003), expressed in the context of communication, the specific language and jargon that is used, the humor that is shared, and the hidden meaning that is implied. C2.1: if the source of information is a strong tie, users are more likely to pay attention to the information from this source on SNS.C2.2: if users are interested in getting to know the source of information better, they are more likely to pay attention to the information from this source on SNS.C2.3: the more overlapping the networks of two users are, the less they are likely to pay attention to the information from each other on SNS.C2.4: the presence of feedback from others in form of ratings and comments will induce users to pay attention to this information on SNS.C2.5: the more frequently users post on SNS themselves, the more they are likely to pay attention to the information that others post.
On the network level of analysis, one can analyze the cohesion of the network vs. availability of structural holes. Each of these network structures can be advantageous for social capital benefits. Network Cohesion: is defined as strongly interconnected ties with each other. Reputation in front of others (common friends can verify information), ensures trusting relationshipsSocial norms and sanctions ensure cooperation and diminishes risk of opportunismStructural Holes: position between two otherwise unconnected clustersMore diverse information Better access to information Advantageous contacts to othersThe more overlapping the networks of users are, the more likely is that the information they possess is redundant and they can also obtain such information from other people to whom they are connected. An increase in mutual friends decreases the value of each additional contact, as the probability that this contact provides new information is relatively low.
Coleman defines cohesion as strongly interconnected ties with each other, already providing evidence for the fact that researchers to some extent equate cohesion and tie strength. Although tie strength and cohesion (network overlap) are correlated, in our study the correlation is only 20%. That is, if one knows someone well, there is a certain probability that they have may mutual contacts. However, tie strength should not be equated with network overlap, as these two measures of network structure may not necessarily coincide.On the one hand, it is possible to imagine highly overlapping networks of two users, who are connected by a weak relationship, such as for example, school classmates. On the other hand, one can be quite close with someone, but the networks may not necessarily overlap, for example two people who live in different cities or belong to different social circles, but had a period of intensive communication at one stage of their lives, such as lovers. In fact, the distribution of responses in our study shows that, out of 80% weak ties, 13% have high network overlap. Out of 20% strong ties, 13% have low network overlap, thus corroborating our propositions.We measure tie strength by subjectively asking them to choose out of a list of friends those whom they know well. This variable is dichotomous, where 1 indicates a strong tie.We measure network cohesion or overlap objectively, by recording the number of mutual friends between the respondent and source of information. Thus, we are able to oppose two different measures of network structure and explore their impact on information value. As we hypothesized, tie strength is positively related to information value, whereas as we saw before, network overlap is negatively associated with it. On the one hand, if the tie is weak, there is low interest in information coming from that person and therefore no motivation to process such information. On the other hand, a high network overlap might result in redundant information and the ability to obtain the same information also from someone else in the network. Therefore, a combination of high tie strength and low network overlap might promise the highest benefits to the users: the diversity of the network allows to get access to the resources that one does not possess oneself, whereas the strong relationship allow to easily obtain those resources if needed.
Main findings: Strong ties: users prefer information from their stronger ties on the network,the persistence of this effect in all of the models we test. Future ties: As stronger ties comprise only a smaller part of the individualsā networks (ca. 20% of all friends whose information they were evaluating), we additionally find that users are also interested in information about their weaker ties that they want to get to know better in the future. By providing constant information updates from these people, SNS environments provide good opportunities to interact more frequently, and as such to develop these relationships. Overlap: users evaluate the information from those friends with whom they have a lot of mutual friends negatively compared to those with whom they have less of them. Thus we empirically confirm the theory of network redundancy proposed by Burt (1992). This is quite an interesting result, as on average mutual friends do not comprise a large part of the userās network (for 60% of the users these are on average just 5%). Moreover, the information about mutual friends is not directly available when users are evaluating information, but only if participants go directly to the profile of a user. Tie strength vs. overlap. That is, considering two ties with similar tie strength, SNS users will be more interested in those users with whom they have less mutual friends. At the same time, considering two users with a similar number of mutual friends, users will be more interested in those with whom they have a stronger relationship. Quadratic effect: We also found that the strength of the negative relationship between network overlap and information value diminishes as network overlap increases. In its turn this implies that past a certain network density, the marginal decrease in information value is negligible. Ratings have postive impact as expected, attract attention Comments are not significant This can be explained by the dual impact of comments on user attention: on the one hand, they might attract the attention of the user to the information that is shared, although on the other hand might create information overload Experience: the more one posts oneself, the more one is interested in the information what others post. Robustness Checks: In terms of robustness checks the specifications have been estimated with robust standard errors derived through bootstrapping with 1000 repetitions (Wooldridge, 2002), but this didnāt change the results significantly, indicating that the covariance matrix is well specified. This model was also tested with fixed effects and a similar pattern of results could be observed. Omitted variable bias: To further illustrate this point, the specifications have been re-estimated first without network overlap and then without tie strength. As tie strength and network overlap are correlated, if we exclude one of them, then a part of one variable will be included into the impact of the other and therefore we will not be able to discern the impact of each of them ā known as omitted variable bias. If the effects of these variables differ (as is the case with the impact of tie strength and network overlap on information value), this omission might make the coefficients smaller or render them insignificant because it pushes them back to 0. Omitting network overlap: we observe that in the first case the estimated coefficient on tie strength 1 goes from 1.02 to 0.91 (random effects) and 1.04 to 0.94 (fixed effects), while the one on tie strength 2 goes from 0.58 to 0.54 (random effects) and 0.62 to 0.58 (fixed effects). If tie strength is excluded, we observe an even more severe case: the coefficients on network overlap become insignificant in both random and fixed effects models.
Tie strength: determines behavior on SNSis positively related to information value Network overlap: on top of tie strength, is negatively related to information valueshould be distinguished from tie strength Social Information:Is the more important, the weaker is the tie comments negatively, whereas ratings positively impact information valueAlthough the direction of causality requires further studies. Experienceis associated with higher benefitscausality? (better say