Interrupting or not: Exploring the effect of social context
on interrupters’ decision making
Department of Informatics
University of Umeå
Department of Information Science
and Media Studies
University of Bergen
In recent decades technology-induced interruptions
emerged as a key object of study in HCI and CSCW
research, but until recently the social dimension of
interruptions has been relatively neglected. The focus of
existing research on interruptions has been mostly on their
direct effects on the persons whose activities are
interrupted. Arguably, however, it is also necessary to
take into account the “ripple effect” of interruptions, that
is, indirect consequences of interruptions within the social
context of an activity, to properly understand interrupting
behavior and provide advanced technological support for
handling interruptions. This paper reports an empirical
study, in which we examine a set of facets of the social
context of interruptions, which we identified in a previous
conceptual analysis. The results suggest that people do
take into account various facets of the social context when
making decisions about whether or not it is appropriate to
interrupt another person.
Interruptions; social context; collaboration; interpersonal
relation; physical proximity; communication.
ACM Classification Keywords
H.5.m. Information interfaces and presentation (e.g.,
Human Factors; Design; Measurement.
With interactive technologies becoming increasingly
pervasive the interruptions caused by the technologies are
becoming pervasive, as well. Understanding how
interruptions happen, their effect on people and their
activities, as well as possible ways to designing
technological support for handling interruptions are key
research questions in HCI and CSCW [4, 9, 11].
Studies of interruptions have been dealing with a range of
issues, in particular: (a) the occurrence of different types
of interruptions in various everyday contexts (e.g., [2, 5]),
(b) effects of interruptions on interrupted activities, which
effects were typically (but not always, see e.g. ) found
negative [3, 6, 10], and (c) technological solutions, which
can help prevent unwanted interruptions from taking
place, as well as help users recover from interruptions if
they do take place (e.g., [1, 7, 12]).
While conceptual analyses and design explorations of
interruptions have undoubtedly produced a number of
significant results, they are, arguably limited in the sense
that they have mostly focused on the direct effects of
interruptions on the persons whose activities are
interrupted (that is, interruptees). Less attention has been
paid to understanding how people decide whether or not
to interrupt (that is, understanding interrupters). For
instance, studies of availability clues, intended to
minimize interruptions, are mostly concerned with how to
help people provide such clues to others rather than how
to utilize availability clues, that others provide.
With some exceptions (e.g. [12, 13, 14]) the social
dimension of interruptions has been neglected. As we
argue in a previous paper , the “ripple effect” of
interruptions, that is, indirect consequences of
interruptions within the social context of an activity, is
underrepresented in existing research. We argue that it is
critical to take these into account to properly understand
interrupting behavior and to be able to develop advanced
technological support for handling interruptions.
In our previous analysis  we described a variety of
“ripple effects”: from “collateral disruption” (the effect an
interruption directed at one person may have on other
people present, such as a mobile phone ringing during a
concert) to “dropping the ball” (a distraction experienced
by one person causes delays in activities of other
participants in a collaborative activity). We also identified
four relevant facets of the social context:
• interpersonal relation (whether or not there exist
a personal relation between interrupter and
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.
NordiCHI’12, October 14-17, 2012 Copenhagen, Denmark
Copyright © 2012 ACM 978-1-4503-1482-4/12/10… $15.00”
• location (whether or not the interuptee is located
in a context where others could be indirectly
disrupted by an interruption),
• communication (whether or not the interruptee is
involved in communication with others), and
• collaboration (whether or not the interruptee is
involved in collaboration with others).
We argued that these facets of context are likely to have
an effect on how (or if) interruptions are taking place. The
aim of the present paper is to address some of the
limitations of existing HCI and CSCW research into
interruptions by providing empirical evidence about
how/if people take social context into account when
making decisions about whether to interrupt another
person or not. The study seeks to find empirical evidence
regarding the following questions:
• Do people take into account social context when
making a decision about whether to interrupt?
• What is the relative importance of individual
facets of social context when making such
GENERAL METHODOLOGICAL CONSIDERATIONS
Conducting an empirical study of how people make
decisions about interrupting others presents a serious
challenge from a research methodology point of view,
especially if the aim of a study is to investigate a set of
specific facets of social context. Direct observations in
natural settings have the advantage of high external
validity but are problematic because of practical and
ethical constraints. Since the researcher does not have
control over the social context, the specific facets of
interest may never be observed within the timeframe of
the study. In addition, such observations can be difficult
to interpret, since the reasons why people make a decision
may not be obvious for an external observer. Finally,
direct observations in real life contexts can undermine
participants’ privacy and integrity.
Direct observations in artificial settings allow researchers
to model situations, in which phenomena of interest are
likely to occur. However, this method is associated with
low external validity. Knowing that the situation at hand
is not “real” may significantly change the decision-
making process in the participants.
Therefore, an approach to studying how people make
decisions regarding interruptions, especially suitable for
an initial exploratory study, appears to be employing
interviews and questionnaires to capture participants’
opinions about, and real-life experience with, making
such decisions. However, a straightforward approach, that
is, simply asking the participants about their opinions and
preferences may make it hard for the participants to relate
their experience to issues in question.
After considering the concerns mentioned above, we have
adopted scenario-assessment as a method for
investigating the effect of different facets of social context
on making decisions regarding interrupting other people.
The facets used were the ones that were identified in our
previous analysis. We did however decide to change the
terminology for one of them from “location” to “physical
proximity” as this better captures what is actually meant.
We constructed a set of concrete scenarios, each
describing a context in which a participant had to decide
whether or not to interrupt a certain person. The contexts
that we used for this purpose were tax office, bus stop,
library, police station, school and accountant’s office.
Then we produced several variations of each scenario by
systematically emphasizing or de-emphasizing certain
facets of social context. The participants were asked to
assess the probability of trying to establish interaction in
the contexts described by each of the variations.
Twenty-five undergraduate students at a Swedish
university, 16 males and 9 females, between 21 and 46
years old (average age of 26), fluent English speakers,
took part in the study.
The materials used in the study comprised sets of
assessment scenarios, each scenario shown on a separate
sheet of paper. In every scenario an imaginary context
was first described, in which one person was supposed to
interrupt another. Then four different additional
conditions were listed. The conditions represented four
possible combinations of two context facets, each of
which could be expressed at two different levels, High vs.
Low. For instance, the person to be interrupted could be a
personal acquaintance (high level of personal
relationship) or stranger (low level of personal
relationship), and he or she could be engaged in a
collaborative activity with other people (high level of
collaboration) or apparently working alone (low level of
collaboration). The participants were asked to assess each
of the four conditions by assigning a percentage
describing the estimated probability, with which they
would interrupt the person described in the scenario.
By systematically combining six different conditions (all
possible combinations of 4 context facets) and 6 types of
context we produced a pool of 36 assessment scenarios.
During the study each participant was presented with a set
of 6 assessment scenarios. These sets were constructed so
that all 36 scenarios were assessed during the study as a
whole. The order of 6 combinations of context facets was
balanced by using a 6x6 Latin Square design.
Copies of assessment scenario sets were printed, sorted,
and stapled, to ensure the correct presentation order. Sets
were given to students, who were manually filling in the
printed copies when sitting in a classroom. One of the
authors was present throughout the procedure.
The procedure of analyzing the data included the
following steps. A table of the 36 scenarios was created,
each respondent’s estimation of the probability of
interrupting under the four conditions filled in and the
average values for all respondents calculated. As a final
step various calculations were made to look for relative
importance of the four facets. An important part of this
last step was to rank the weight of difference facets. As all
four facets were compared against each other in different
scenarios by the creation of different conditions (high vs.
low, low vs. high etc.) they could easily be ranked
through a grading process. If the average value of making
an interruption under the condition that physical
proximity is high and interpersonal relation is low
exceeds the average value for the opposite, then physical
proximity has more weight than interpersonal relation. As
all facets were tested against each other, their weights
were measured by giving them a value for every time they
were considered superior and when summarizing all
values the relative weight of all facets were established.
The results show that respondents do take the social
context into consideration when deciding whether or not
to interrupt another person. In all 36 scenarios it is shown
that respondents estimate the probability of making an
interruptions as lower if there exist no previous relation,
the interruptee is located in a context where others could
be disturbed by an interruption, or is involved in
communication or collaboration with others. The
difference, for all scenarios, between the condition where
the facets are assumed to speak in favor of an interruption
(i.e. when an interpersonal relation exists, there are few or
no bystanders that could be disturbed and the interruptee
is not involved in any communication or collaboration)
and the opposite, ranges from 47% to 63%.
The degree of influence that the social context has on a
decision about whether or not to interrupt does however
differ between scenarios and combinations of facets. A
scenario that takes place in an accountant’s office where
the facets interpersonal relation and physical proximity
(of other people) are combined in four different
conditions shows the largest difference between estimated
probabilities of interruption for different conditions
(89%). More specifically, according to estimations of the
probability of making an interruption, it is 89% more
likely that an interruption would occur under the
condition that there exist a previous relation and there are
no other people on the scene, than if it were the other way
around. The scenario that showed the lowest difference in
estimated probability of making an interruption between
different conditions is the one that takes place in a tax
office and the facets of communication and physical
proximity are combined. In that case there was a
difference, but as low as ≈13%. When comparing the
estimated probability of making an interruption under the
same conditions but in different scenarios it becomes
evident that although the estimations are consistent, the
range in percentage varies. When, for example, a
combination of interpersonal relation and involvement in
communication is assessed in the “bus stop” scenario and
the “library” one, the difference is as high as 47,8%.
The design of the study ensured that all facets were
combined and tested in all conditions, which allows us to
make inferences about facets that have more weight than
the others. The results show that physical proximity is the
facet with most weight (i.e. if there are other people
nearby that could be disturbed by the interruption), an
interruptee’s involvement in communication the second,
interpersonal relation the third and interruptees’
involvement in collaboration the least dominant. Worth
mentioning is however that the difference between
physical proximity and communication is only ≈13%.
Other interesting results are that in some scenarios the
strength relationships between different facets differ from
the overall pattern presented above. For example in the
tax office scenario the interruptee’s involvement in
collaboration with others, even if with a small margin,
outweigh an existing previous relation (with 1,7%), or in
the scenario that takes place at a school where
involvement in communication outweigh physical
proximity (with 6%).
This paper continues our previous work on social
dimensions of interruptions (see ) by presenting
empirical evidence of their existence and importance. It
complements work on interruptions by showing how
people take social contexts into consideration before
interrupting others and also by comparing the level of
influence of different facets of these dimensions.
Exploring interruptions from the perspective of the
interrupter, and not only the interruptee, is, in our opinion,
a necessary step towards an improved understanding of
the phenomena and developing more advanced
technological support for interruption handling.
One of the main challenges for HCI and CSCW research
into interruptions is finding novel technological solutions
that would simultaneously address different, potentially
conflicting concerns. On the one hand, the more
information about interruptee’s current social context is
provided to the interrupter, the easier it is for the latter to
decide whether or not to interrupt. On the other hand,
providing such information may undermine interruptee’s
privacy. Understanding how exactly people make
interruption decisions can help identify ways to balance
these concerns, that is, provide enough information to
make a decision without revealing too much. The findings
of the study reported in this paper allow us to make to
some tentative conclusions about the facets of social
context taken into account when making decisions about
Even though the study only included a limited number of
respondents, the results suggest, in each and every
scenario, that people are more likely to make an
interruption if there is a previous personal relationship,
there is little risk for disturbing other people except the
interruptee, and that the interruptee is not involved in
communication or collaboration with others. Although
there are some exceptions in some scenarios, the overall
picture shows how some facets are considered as more
important to take into consideration before making an
interruption than others. As mentioned above, whether or
not there are other people around that might be disturbed
by an interruption is experienced as more important than
the other investigated facets. Worth mentioning however
is that the facet that was shown to be least influential,
involvement in collaboration, still had a significant impact
on the estimated probability of initiating an interruption
by the respondents.
Another important observation is that different contexts
(e.g. a police station, library or tax office) have a clear
effect on the estimated probability of making
interruptions under similar conditions. This could be
caused by how the scenarios were described, but it is also
likely that respondents co-created the scenarios by adding
their own experiences, norms and understandings to these
descriptions. This could at least partly explain individual
differences found in the empirical data, differences that
are partly hidden as a result of our analysis.
Even though it is tempting to consider immediate
implications of the findings presented in this paper for
design and evaluation of interactive technologies, much
more work is needed. It should be established whether the
findings could be generalized to a wider population, as
well as to technology-mediated communication and
collaboration. We are preparing to conduct another set of
scenario-assessments to include additional scenarios and a
far higher number of respondents with more diverse
characteristics in terms of occupation and age. This will
further improve the validity of our claims regarding social
contexts and interruptions.
1. Altmann E. M., and Trafton J. G. Task interruption:
Resumption lag and the role of cues. In Proc. CogSci
2004, Lawrence Erlbaum Associates (2004), 43-48.
2. Ammons S. K., and Markham W. T. Working at
home: Experiences of skilled white collar workers,
Sociological Spectrum, 24, 2 (2004), 191-238.
3. Bailey, B. P., Konstan, J. A., and Carlis, J. V. The
effects of interruptions on task performance,
annoyance, and anxiety in the user interface. In Proc.
INTERACT 2001, IOS Press (2001), 593-601.
4. Burmistrov, I. and Leonova, A. Do interrupted users
work faster or slower? The micro-analysis of
computerized text editing task. In Proc. of HCI
International 2003, Lawrence Erlbaum Associates
5. Czerwinski, M., Horvitz, E., and Wilhite, S. A diary
study of task switching and interruptions. In Proc.
CHI 2004, ACM Press (2004), 175-182.
6. Dodhia, R. M., and Dismukes, R. K. Interruptions
create prospective memory tasks. Applied Cognitive
Psychology, 23, 1 (2009), 73-89.
7. Hameed, S., Ferris, T., Jayaraman, S., and Sarter, N.
Using informative peripheral visual and tactile cues to
support task and interruption management. Human
Factors, 51, 2 (2009), 126-135.
8. Harr, R., and Kaptelinin, V. Unpacking the Social
Dimension of External Interruptions. In Proc. GROUP
2007, ACM Press (2007), 399-408.
9. Iqbal, S. T. and Bailey, B. P. Oasis: A framework for
linking notification delivery to the perceptual structure
of goal-directed tasks. ACM Transactions on
Computer-Human Interaction (TOCHI), 17, 4 (2010),
10. Mark, G., Gudith, D., and Klocke, U. The cost of
interrupted work: More speed and stress. In Proc. CHI
2008, ACM Press (2008), 107-110.
11. Perlow, L. The time famine: Toward a sociology of
work time. Administrative Science Quarterly, 44, 1
12. Petersen, S. A., Cassens, J., Kofod-Petersen, A., and
Divitini, M. To be or not to be aware: Reducing
interruptions in pervasive awareness systems. In Proc.
UBICOMM 2008, IEEE Computer Society (2008),
13. Ritterskamp, C. The Collaborative Nature of
Interruption Handling. In Proc. HICSS 2011, IEEE
Computer Society (2011), 1-10.
14. Tolmie, P., Crabtree, A., Rodden, T., and Benford, S.
“Are You Watching This Film or What?” Interruption
and the Juggling of Cohorts. In Proc. CSCW 2008,
ACM Press (2008), 257-266.
15. Zijlstra, F. R. H., Roe, R. A., Leonova, A. B., and
Krediet, I. Temporal factors in mental work: Effects of
interrupted activities. Journal of Occupational and
Organizational Psychology, 72, 2 (1999), 163-185.