Predicting Tie Strength With Social Media
Eric Gilbert and Karrie Karahalios
University of Illinois at Urbana-Champaign
[egilber2, kkarahal]@cs.uiuc.edu
ABSTRACT
Social media treats all users the same: trusted friend or total
stranger, with little or nothing in between. In reality, rela-
tionships fall everywhere along this spectrum, a topic social
science has investigated for decades under the theme of tie
strength. Our work bridges this gap between theory and
practice. In this paper, we present a predictive model that
maps social media data to tie strength. The model builds on
a dataset of over 2,000 social media ties and performs quite
well, distinguishing between strong and weak ties with over
85% accuracy. We complement these quantitative findings
with interviews that unpack the relationships we could not
predict. The paper concludes by illustrating how modeling
tie strength can improve social media design elements, in-
cluding privacy controls, message routing, friend introduc-
tions and information prioritization.
Author Keywords
Social media, social networks, relationship modeling, ties,
sns, tie strength
ACM Classification Keywords
H5.3. Group and Organization Interfaces; Asynchronous
interaction; Web-based interaction.
INTRODUCTION
Relationships make social media social. Yet, different rela-
tionships play different roles. Consider the recent practice
of substituting social media friends for traditional job refer-
ences. As one hiring manager remarked, by using social
media “you’ve opened up your rolodex for the whole world
to see” [38]. To the dismay of applicants, employers some-
times cold call social media friends expecting a job refer-
ence “only to find that you were just drinking buddies.”
Although clearly not the norm, the story illustrates a basic
fact: not all relationships are created equal.
For decades, social science has made much the same case,
documenting how different types of relationships impact
individuals and organizations [16]. In this line of research,
relationships are measured in the currency of tie strength
[17]. Loose acquaintances, known as weak ties, can help a
friend generate creative ideas [4] or find a job [18]. They
also expedite the transfer of knowledge across workgroups
[20]. Trusted friends and family, called strong ties, can af-
fect emotional health [36] and often join together to lead
organizations through times of crisis [24]. Despite many
compelling findings along this line of research, social me-
dia does not incorporate tie strength or its lessons. Instead,
all users are the same: friend or stranger, with little or noth-
ing in between. Most empirical work examining large-scale
social phenomena follows suit. A link between actors either
exists or not, with the relationship having few properties of
its own [1, 2, 27].
This paper aims to bridge the gap, merging the theory be-
hind tie strength with the data behind social media. We ad-
dress one central question. With theory as a guide, can so-
cial media data predict tie strength? This is more than a
methodological or theoretical point; a model of tie strength
has the potential to significantly impact social media users.
Consider automatically allowing the friends of strong ties to
access your profile. Or, as one participant cleverly sug-
gested, remaking Facebook’s Newsfeed to get rid of “peo-
ple from high school I don't give a crap about.” The model
we present builds on a dataset of over 2,000 Facebook
friendships, each assessed for tie strength and described by
more than 70 numeric indicators. It performs with surpris-
ing accuracy, modeling tie strength to 10-point resolution
and correctly classifying friends as strong or weak ties more
than 85% of the time.
We begin by reviewing the principles behind tie strength,
and then discuss its proposed dimensions. Using theory to
guide the selection of predictive variables, we next present
the construction of our tie strength model. It performs well,
but not perfectly. To understand our model’s limitations, we
also present the results of follow-up interviews about the
friendships we had the most difficulty predicting. The paper
concludes by applying our findings toward implications for
theory and practice.
TIE STRENGTH
Mark Granovetter introduced the concept of tie strength in
his landmark 1973 paper “The Strength of Weak Ties” [17].
In this section we review tie strength and the substantial
line of research into its characteristics. We then discuss four
researchers’ proposals for the dimensions of tie strength,
laying a foundation for our treatment of it as a predictable
quantity. The section concludes by introducing the research
questions that guide the rest of this paper.
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies
bear this notice and the full citation on the first page. To copy otherwise,
or republish, to post on servers or to redistribute to lists, requires prior
specific permission and/or a fee.
CHI 2009, April 4–9, 2009, Boston, Massachusetts, USA.
Copyright 2009 ACM 978-1-60558-246-7/09/04...$5.00.
CHI 2009 ~ Online Relationships April 6th, 2009 ~ Boston, MA, USA
211

Definition and Impact
The strength of a tie is a (probably linear) combination of the
amount of time, the emotional intensity, the intimacy (mutual
confiding), and the reciprocal services which characterize the
tie. [17]
While Granovetter left the precise definition of tie strength
to future work, he did characterize two types of ties, strong
and weak. Strong ties are the people you really trust, people
whose social circles tightly overlap with your own. Often,
they are also the people most like you. The young, the
highly educated and the metropolitan tend to have diverse
networks of strong ties [31]. Weak ties, conversely, are
merely acquaintances. Weak ties often provide access to
novel information, information not circulating in the closely
knit network of strong ties.
Many researchers have adopted tie strength as an analytic
framework for studying individuals and organizations [16].
(Google Scholar, for instance, claims that over 7,000 papers
cite “The Strength of Weak Ties” [15].) The social support
offered by strong ties can actually improve mental health
[36]. Banks that find the right mix of weak and strong ties
to other firms tend to get better financial deals [39]. It has
also been shown that weak ties, as opposed to strong ones,
benefit job-seekers [18]. However, socioeconomic class
reverses this effect: job-seekers from lower socioeconomic
backgrounds often rely heavily on strong ties [16].
Strong ties between employees from different organiza-
tional subunits can help an organization withstand a time of
crisis [24]. Yet, strongly tied coworkers are also the ones
likely to create crises by pushing for institutional change
[23]. Employees who weakly tie themselves beyond organ-
izational boundaries tend to receive better performance re-
views and generate more creative ideas [4]. Weak ties also
act as a conduit for useful information in computer-
mediated communication [8]. However, weak ties often rely
on a few commonly available media [22], whereas strong
ties diversify, communicating through many channels [21].
The Dimensions of Tie Strength
At what point is a tie to be considered weak? This is not sim-
ply a question for the methodologically curious … the theory
makes a curvilinear prediction. How do we know where we
are on this theoretical curve? Do all four indicators count
equally toward tie strength? [23]
Granovetter proposed four tie strength dimensions: amount
of time, intimacy, intensity and reciprocal services. Subse-
quent research has expanded the list. Ronald Burt proposed
that structural factors shape tie strength, factors like net-
work topology and informal social circles [5]. Wellman and
Wortley argue that providing emotional support, such as
offering advice on family problems, indicates a stronger tie
[40]. Nan Lin, et al., show that social distance, embodied by
factors such as socioeconomic status, education level, po-
litical affiliation, race and gender, influences tie strength
[29].
In theory, tie strength has at least seven dimensions and
many manifestations. In practice, relatively simple proxies
have substituted for it: communication reciprocity [11],
possessing at least one mutual friend [37], recency of com-
munication [28] and interaction frequency [13, 17]. In a
1984 study, Peter Marsden used survey data from three met-
ropolitan areas to precisely unpack the predictors of tie
strength [33]. While quite useful, Marsden pointed out a
key limitation of his work: the survey asked participants to
recall only their three closest friends along with less than
ten characteristics of the friendship.
The present research can be seen as updating Marsden’s
work for the era of social media. Our work differs primarily
in setting and scale. By leveraging social media, partici-
pants no longer have to recall; we can take advantage of
long friend lists and rich interaction histories. In this way,
our work also overcomes the problem of retrospective in-
formant accuracy [3, 30, 32]. In addition, a tie strength
model built from social media has the potential to feed back
into social media, in ways that benefit its users.
Figure 1. The questions used to assess tie strength, embedded into a friend’s profile as participants experienced them. An auto-
mated script guided participants through a random subset of their Facebook friends. As participants answered each question by
dragging a slider, the script collected data describing the friendship. The questions reflect a diversity of views on tie strength.
CHI 2009 ~ Online Relationships April 6th, 2009 ~ Boston, MA, USA
212