Email Duration, Batching and Self-interruption:
Patterns of Email Use on Productivity and Stress
Gloria Mark1, Shamsi T. Iqbal2, Mary Czerwinski2, Paul Johns2, Akane Sano3, Yuliya Lutchyn2
1Department of Informatics
University of California, Irvine
One Microsoft Way
Massachusetts Institute of Technology
Irvine, CA 92697 USA
Redmond, WA 98052 USA
Cambridge, MA 02139 USA
While email provides numerous benefits in the workplace,
it is unclear how patterns of email use might affect key
workplace indicators of productivity and stress. We
investigate how three email use patterns: duration,
interruption habit, and batching, relate to perceived
workplace productivity and stress. We tracked email usage
with computer logging, biosensors and daily surveys for 40
information workers in their in situ workplace
environments for 12 workdays. We found that the longer
daily time spent on email, the lower was perceived
productivity and the higher the measured stress. People who
primarily check email through self-interruptions report
higher productivity with longer email duration compared to
those who rely on notifications. Batching email is
associated with higher rated productivity with longer email
duration, but despite widespread claims, we found no
evidence that batching email leads to lower stress. We
discuss the implications of our results for improving
organizational email practices.
Email; sensors; productivity; workplace; stress;
interruptions; in situ study
ACM Classification Keywords
H.5.3 [Information Interfaces and Presentation (e.g., HCI)]:
Group and Organization Interfaces; K.4.m [Computers and
How do patterns of email use affect the workplace
experience? In today’s information driven world, email
continues to be a ubiquitous communication medium on
both organizational and personal levels [5, 10, 40].
Communication in corporate organizations happens mostly
through email . Email has been shown to be very useful
for assigning and communicating to do’s , for
coordinating and assigning tasks amongst colleagues ,
for task management and archiving information , and
for storing, retrieving and sharing information easily .
However, it is well established by numerous studies that
email leads people to feel cognitively overloaded, e.g. [2, 3,
9, 41]. The popular press has documented this concern: a
search in the Google newspaper archives has produced over
166,000 news articles on the sole topic of email overload in
the workplace. Sherry Turkle reflects this sentiment as “we
don’t do email, our email does us” . Research studies
have documented concerns from users about the challenge
of keeping up with email [3, 9, 48]. While having good
organizational skills can facilitate email management ,
such skills are not universal and their lack may lead to a
number of negative outcomes.
Studies on email management practices in the workplace
have shown that the time employees spend in managing
email comprises a significant portion of their daily
activities. A 2012 report from the McKinsey Global
institute reveals that 28% of an employee’s workweek is
spent on reading, composing or responding to email .
Also, given the culture of reliance on email for information
exchange in organizations, people also tend to frequently
check email, either triggered by notifications or self-
interruptions . Yet it remains an open question how the
effects of extensive engagement in email interactions
affects people’s workplace experience. In particular, the
relationship between email usage, productivity, and stress
in the workplace is complex and not well explored in the
literature due to its challenging nature.
In this paper, we explore email usage, in terms of how
people check email (self-interrupting or by using
notifications), the time spent on email, and temporal
patterns of checking email (batching or continual), and how
it affects productivity and stress. While many studies have
typically relied on self-reports of email usage, e.g. ,
research shows that such subjective measures grossly
overestimate the time spent using information technology
. To obtain a more reliable measure of participants’
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. Copyrights for
components of this work owned by others than the author(s) must be
honored. Abstracting with credit is permitted. To copy otherwise, or
republish, to post on servers or to redistribute to lists, requires prior
specific permission and/or a fee. Request permissions from
CHI'16, May 07 - 12, 2016, San Jose, CA, USA
Copyright is held by the owner/author(s). Publication rights licensed to
email use, we continuously logged our participants’
computer activity as they conducted their normal work
tasks. Our research questions called for using varied
methods, so we combined computer activity logging with
physiological data and daily surveys, where we measured
affective and cognitive parameters of 40 information
workers for 12 days in their in situ workplace
environments. This research is part of a larger project,
HealthSense, to study the wellbeing of information workers
in the workplace.
Our findings show that some patterns of email use are
associated with lower perceived productivity and higher
stress. The longer daily duration spent on email, the lower
the assessed productivity and the higher the stress. With
high email use, people who chose when to self-interrupt to
deal with email, and "Batchers", people who cluster email
use, assessed their productivity higher at the end of the day
compared to those who check email triggered by email
notifications, and to those who check email consistently. To
our knowledge, our study is the first in situ multi-method
investigation of email activity, workplace outcomes and
stress. Our results lay ground for future theoretical
exploration of these effects, and provide valuable practical
lessons for organizations and knowledge workers.
Email usage in the workplace: an overview
Research suggests that people spend quite a bit of time
checking their email daily. Studies found that users check
their email around 11 times per hour , that 84% of users
keep their email up in the background at all times, and 64%
of users used notifications to access email at least some of
the time . Czerwinski et al.  found that email
accounted for 24% of the tasks information workers
reported performing in a daily diary study. Fisher et al. 
reported an average of 87 emails received per day, while
Mark et al.  found that users in a logging study spent an
average of 34.5 minutes per day on email. Jackson et al.
 discovered that 70% of all emails received were
opened within 6 seconds of their receipt, and it took an
average of 64 seconds to resume the task interrupted by
email. Obviously, these disparate estimates could be due to
a variety of factors, such as culture, workplace, and
measurement technique. The bottom line is that people are
using email quite a lot, which in turn could have a variety
Benefits of email in the workplace
Multiple studies have shown that continual email
engagement is not unwarranted: email provides many
benefits in the workplace [3, 47]. As such, it is not really an
option for users to totally “opt out”, though this has been
shown to be beneficial for reducing stress . Mano and
Mesch  found email to be helpful in speeding up
communication and benefiting performance in the
workplace. Email supports both information management
and communication . So email certainly has been
shown repeatedly to be a multifaceted tool potentially
benefiting workplace productivity.
Cost of the ubiquity of email in the workplace
Despite its usefulness, research does show that the ubiquity
of email has its costs. A number of factors have been
identified that contribute to the feeling of email causing
cognitive overload, including a lack of clarity of email
requests , the work being demanded in the emails ,
poor email management strategies , a loss of control ,
problems keeping track of email threads , interruptions
due to email , and social pressure to respond (quickly),
especially if the sender is higher up in the organizational
hierarchy [2, 41]. Email generally imposes more costs on
the recipient than the sender, especially when information is
requested or when work is delegated . Nevertheless,
despite the purported costs, email still remains a primary
communication mode in the workplace.
Effect of email on productivity
Because of the benefits offered by email in the workplace
as well as the corresponding costs, the relationship between
email use and productivity is complex. As such, few studies
have addressed the relationship of email use to productivity.
A broad measure of communication technology overload
was found to negatively correlate with productivity  as
did a more specific measure of email overload . On the
other hand, the number of email messages received
increased perceived workplace effectiveness , even
though research shows only about 30% of received email
requires action . Also, since 32% of emails remain
unread , this raises the question of what other aspects of
email use might affect productivity.
Loss of productivity with email use has been explained as
due to the time spent continually monitoring email, taking
time away from other activities . People have reported
being lost in email 23% of the time, often due to diversions
, which could increase their time on email without
feeling productive. While these studies provide insight into
the different ways people interact with email, there have
been no studies quantifying how different email usage
behaviors might affect productivity.
Effect of email on stress
Studies show that email usage is, indeed, negatively related
to stress . In one study users were asked to turn off their
email for a week while they wore heart rate monitors to
measure heart rate variability (HRV, a validated measure of
stress/depression) . Compared to a baseline period with
email use, the HRV signals revealed less stress when email
was turned off, even though other communication channels
like the phone, instant messaging, etc., were still available
to be used. In another study, when participants were
instructed to limit the frequency of checking their email,
they experienced less stress . People have also reported
anxiety in not being able to keep up with their inbox, which
could result in missing critical information .
Some studies have argued that time spent on email creates
additional work for the user which in turn elevates stress
. Though without empirical support, these claims are
based on the idea that time spent on email creates more
add-on work for people due to its affordances.
Communication is easier and faster via email than written
notes and thus it creates more messages that people must
spend time with, not only in responding to them, but also in
organizing and filing . Also, as it is easy for the sender to
make requests and delegate work , this creates new tasks
which the recipient may not view as critical to work--some
of which must be conducted through email . Email
creates interruptions which involve extra work for users to
reorient back to the task at hand , and which could lead
to stress. In one study email was the only communication
tool to which stress was attributed .
Other claims are that stress is due to the time spent on email
which extends the workday  and to the volume of email
received : a positive correlation was found with time
spent on email and number of incoming emails .
However, it could be not just the duration of time alone that
has an effect on the workplace experience but the email
management strategies that people employ. The research
streams of email duration, interruptions, and email overload
have not been well linked together to understand how
experience with email affects the workplace experience.
Further, studies have either been in the laboratory, done
with surveys, or in situ without the use of objective
measures of stress. Our work builds on previous studies
relating email overload to lower productivity [22, 37].
Whereas communication and email overload have been
examined, we look at particular email usage patterns and
how they might affect productivity. Similarly, our study
also builds on the work relating email to stress [23, 28]. No
one has examined how duration of time on email might
affect stress and whether strategies of self-interruptions to
check email or batching email could reduce stress.
Despite the documented studies of email overload, it is
important to consider that email not only increases the
incoming stream of information and tasks, but also provides
more structured support for communication and
coordination, which may be vital to accomplishing tasks
related to work. One reason for feeling overloaded from
information can be attributed to when the demands on time
to deal with information are greater than the amount of time
available, cf . Investing time to manage email takes
time away from other activities. Interruptions from email
were found to take time from other more crucial tasks in the
workplace . Thus, dealing with email could lead people
to feel that they are compromising engagement in other
types of work which could be more productive for them. On
the other hand, as a large proportion of email use concerns
task management , it might be expected that work on
email could lead to a sense of increased productivity. Often
tasks originate in email  and dealing with email could be
a way of accomplishing tasks.
Thus, it is an open question how specifically different
patterns of email use might correlate with productivity and
stress. Based on a review of the literature, we selected the
following email usage patterns to examine: duration of time
on email, types of interruptions, and temporal patterns of
Email duration. As email use comprises a significant
portion of the day, we feel that a measure of email duration
is important to examine. Yet the few studies that have
looked at the effects of spending time on email have found
contradictory results. In a year-long study of college
students, hours of email use per week were negatively
associated with stress . Yet workplace studies found
that the amount of time employees spent on email was
positively correlated with feeling overloaded [2, 39].
Except for the study of Bradley et al. , who found no
relation of email duration with stress, these studies involved
self-reports which have been found to inaccurately reflect
actual time with computer usage .
Yet with the exception of the study of Barley et al.,  who
also used self-reports, studies have not directly measured
the relationship of time spent on email and its effect on
productivity and stress. We find this surprising, as a fair
amount of research documents that email comprises a
significant portion of the day [8, 14, 19, 26, 35]. A large
amount of research also has addressed reasons for email
overload, e.g., [2, 3, 9, 10, 11, 47, 48]. Yet these streams of
research have not been well linked together and there is a
lack of research using objective measures of duration of
time spent on email to examine its association with
productivity and stress in the workplace. We examine
whether the amount of time spent on email is associated
with productivity and stress.
Interruption types. We examine how a person's habits of
email checking, triggered primarily by email notifications
or by self-interruptions, might affect productivity.
Interruptions, documented to be disruptive in work and
requiring a recovery time, e.g., [30, 38, 15], could have an
impact on the workplace experience. People can check
email in different ways: by primarily relying on
notifications or by primarily checking on one's own (and
not waiting for, relying on, or reacting to notifications), or
using both strategies. These different strategies may be
associated with different productivity or stress levels in the
workplace. For example, if one primarily checks email on
their own, this could reflect better coordination of time,
leading one to feel more productive. There may also be an
interaction with one's interruption habit for checking email
and the amount of time one spends on email. Whereas the
time involved in dealing with messages (reading,
responding, filing, etc.) could relate to productivity, it is an
open question whether relying on email notifications or
checking on one's own might be associated with
productivity and stress.
Batching behavior. Restricting email use to certain times of
the day has been presented as a solution for email
management [28, 36]. The popular media is abound with
claims that using email at set times during the day will
reduce stress and increase productivity, e.g. . The
argument for restricted use, termed "batching email", is that
setting aside times to do email should reduce interruptions,
leaving the rest of the day to focus on other work. This
could potentially increase perceived productivity and
reduce stress. In a study where people adopted a once-a-day
email strategy, their time on email was significantly
reduced though it did not affect stress . Another study
that asked people to restrict their email use to set times did
find though that it lowered stress . Yet in a survey study
of email use, it was found that checking email as it arrives
was associated with lower cognitive load compared to
checking email at defined times . The claims about
batching email, though widespread, remain unsupported
due to conflicting results. In this research question, we will
first investigate whether we can identify profiles that
characterize whether people primarily restrict email use to
certain times of the day or rather use it continually
throughout the day. We will then examine whether batching
email is associated with productivity and stress.
Our research questions on workplace productivity are:
RQ1a. How is time spent on email associated with assessed
productivity in the workplace?
RQ1b. How is interruption type, primarily checking email
triggered by notifications or by self, associated with
assessed productivity in the workplace?
RQ1c. How is batching email associated with assessed
productivity in the workplace?
Our research questions related to workplace stress are:
RQ2a. How is time spent on email associated with stress in
RQ2b. How is interruption type, primarily checking email
triggered by notifications or by self, associated with stress
in the workplace?
RQ2c. How is batching email associated with stress in the
Procedure and participants
We conducted an in situ study with 40 participants (20
females, 20 males). Participants were volunteers working in
a research division of a large corporation, and worked in
different job roles: administrative support, engineering, and
management. Participants gave informed consent and were
compensated with a $250 gift card.
Participants were asked to be in the study for 10 full
business days; however due to technical problems or
scheduling issues, participants averaged 12 study days.
During the study period, physiological data to measure
stress was collected from a heart rate monitor worn around
the chest during all waking hours. Computer activity at
work was logged during all business hours. Prior to the
beginning of the study, we met with participants
individually to explain the study procedure, install the
software, and to instruct them on how to use the heart rate
monitors. Participants were instructed to work as they
normally would throughout the workday. In addition, we
administered a pre-study survey with a number of
demographic, work, and stress measures. Participants were
also sent a daily evening questionnaire, where they reported
their perceived productivity for that day.
All volunteers were assured that their data would be kept
private and aggregated, that no content would be associated
with their information, and that they would remain
anonymous. Upon completion of the study, one of the
researchers interviewed all the participants to confirm that
they followed the study protocol as instructed, and to learn
about any unusual circumstances that could have had an
effect on the data provided by the participants.
Table 1 shows a summary of measures, detailed as follows.
Email Duration Proportion was measured as the ratio of the
time spent on email interactions and total time spent on
computer interaction. We normalized this measure per
person. Time spent on email was logged automatically via
custom-built Windows Activity Logging software. This
logging software tracks every open application, which
window is in the foreground, and whether the user is
interacting with that window (with mouse, keyboard, touch,
etc.). We measured the total duration of email client use.
Email duration was defined as the number of seconds that
the email client was in the foreground window, ending
when the user either changed windows or the computer had
no keyboard or mouse activity for a period of five minutes.
As participants at times might not be using their computer
for various reasons (e.g., they might be at a meeting), we
used only those hours of data when the computer was used
(i.e., the logging data showed that computer duration was
greater than zero for that hour).
Interruption Type was measured in the post-study interview
by the following question: I check email: 1) Always when
triggered by an external notification and never on my own;
2) Much more often when triggered by an external
notification than on my own; 3) About half the time when
triggered by an external notification, half the time on my
own; 4) Much more often on my own than when triggered
by an external notification, 5) always on my own and never
when triggered by an external notification, and 6) I don't
have email notifications. The Interruption Type measure
was categorized into two levels: responses 1 and 2 were
combined into "External interruptions" (External), and
responses 4, 5, and 6 were combined into a measure of
"Self interruptions" (Self). Participants who gave response
3 (half external/half self) were not used in the analysis.
Batching behavior. In detailed visual inspection of plots of
the data for each participant, of their email duration over
the day, we noticed distinct patterns. Some participants
tended to cluster or "batch" their email use, usually in 2 or 3
times per day. Others showed a distinct pattern of checking
email more or less continuously throughout the day. Still a
third group showed a mixed strategy, where sometimes they
would "batch" their email and sometimes check it
continually. We divided participants into different groups as
follows. We first calculated, for each hour of work that day,
the percentage of their total daily email use. For example, if
a person used email continually throughout the day, and
worked 10 hours, then the expected value of email duration
for each hour should be 10% of the total. For an 8-hour day,
the expected value would be 12.5% of the total. Based on
our inspection, observing that many people had 3 peaks of
usage, we took the sum of the 3 highest hours of email
duration for each person, normalized it by the hours of
work that day, and then calculated what proportion of total
daily email was done in those hours. For example, for a 10-
hour day, we would expect that the proportion in 3 hours
would be 30% if a person checked email evenly. But if they
batched email, they may use 60% of their total email
duration in 3 hours. Based on carefully inspecting the
distribution of the entire sample, we used a cut-off criteria
of selecting participants whose 3 highest hours of email use
comprised 50% or more of their total email use. We used
this criteria to create a user Profile of "Batchers". Based on
a visual coding of the data by two independent coders, the
rest of the participants were coded as "Consistents" (where
email use was fairly consistent over the workday), and
"Mixed strategy" where users used a mix of batching and
consistent strategies. The two independent coders were in
100% agreement. The profile counts were as follows:
Batchers: 11, Consistents: 23, Mixed: 6. Figures 1a and 1b
illustrate profiles of a "Batcher" and a "Consistent" checker.
We thus created three Profiles of users: "Batchers", whose
pattern of email use was to cluster email use in two or three
hour periods, "Consistents", whose pattern of email use was
fairly consistent throughout the day, and "Mixed", whose
email use was a mixed strategy.
Email Checks was measured as the number of separate
times that the email client switched to the foreground.
Productivity. In information work, an objective measure of
productivity is difficult to obtain. Performance reviews (e.g.
bi-annual) could provide a measure but these are not fine-
grained enough to look at the relationship with daily email
use. We constructed an index of productivity using six
dimensions included in the daily end of day survey: "How
much did you accomplish today based on what you had
planned to accomplish?”, “How efficient do you feel you
were today in performing your work?”, "How satisfied
were you in what you accomplished today?", "How
effectively do you feel you managed your time today?",
"How would you evaluate the quality of the work you did
today?", and “Overall, how productive do you feel you
were today?”. All responses were measured on a 7-point
Likert scale, with 1=not at all, and 7=extremely. The item
dimensions were highly correlated (with correlations
ranging from .68 to .94), so we combined them additively
to construct an index measure of Productivity.
Stress level was determined from the continuous stream of
cardiovascular data measured by digital heart rate monitors
that participants wore during all waking hours for the entire
duration of the study. We used the Zephyr HXM BT
(bluetooth) heart rate monitor. A custom-built mobile phone
application pulled the data from the Zephyr Heart Monitor,
and uploaded that data into Azure cloud storage. Stress was
estimated based on heart rate variability (HRV) – a well-
validated indicator of mental stress that is used extensively
in research and clinical studies (see  for a review). HRV
is a measure of variations in intervals between consecutive
heartbeats. We used the RMSSD (root mean square of
successive differences) as a measure for calculating HRV
(see ). Perhaps counter-intuitively, the relationship is
inverse, so that the lower the RMSSD measure, the higher
the amount of stress, as the body is regulating itself through
the sympathetic nervous system. Stress was measured to the
Figure 1a. Data of a user who batches email use. Y-axis
shows percentage of daily email done in that hour
Figure 1b. Data of a user who consistently checks email.
second, and then for each hour we computed the average
level of stress for that hour.
The RMSSD was computed each second based on the
variance over the prior 5 minutes. For each hour then, we
compared the average RMSSD along with email duration
and number of email checks. HRV has been used in other in
situ empirical studies, e.g. .
We used the following control variables in our study.
Job characteristics. Email is a communication tool, and an
employee’s job role may significantly affect the amount and
dynamics of its usage. For example, a person with
administrative support duties may process hundreds of
messages a day and have his email client constantly in the
foreground of the computer screen, whereas an engineer
may have her email closed, and only check email during
short, scheduled breaks. To control for such differences, we
took into account our participants’ job roles. Instead of
using a rather broad taxonomy of job titles, we relied on
two fundamental dimensions suggested by Karasek in his
Job Content Questionnaire: job demands, and job decision
latitude . Job demands is an index measure computed
from five items such as “My job requires working very
fast”, “I am not asked to do an excessive amount of work”
(1=strongly disagree, 4=strongly agree). Job decision
latitude is the cumulative measure of an employee’s skill
discretion and decision-making authority, measured by nine
items such as “My job requires a high level of skill” and “I
have a lot to say about what happens on my job”.
Participants answered these questions in the general survey.
Productivity Software duration was measured based on the
logging data. It is possible that a person's productivity
assessment could be based on the amount of time that is
spent using software that supports task features: e.g. writing
documents, doing analyses, creating presentations, or
coding software. We thus controlled for the daily duration
of productivity software use. We measured the total
duration of use of applications that were coded as
"productivity software": the most commonly used
applications in this category were Word, Excel, Powerpoint,
Visual Studio, Matlab, and OneNote. Productivity software
duration was defined as the number of seconds that these
applications were in the foreground window, ending when
the user either changed windows or the computer had no
keyboard or mouse activity for a period of five minutes.
Baseline stress was measured in the pre-study survey based
on the Perceived Stress Scale (PSS) . Because we are
measuring fluctuating stress (with HRV), we used the PSS
score as a baseline measure of stress to control for, as it
measures a global level of stress. The PSS consists of 14
items and measures an individual's subjective evaluation of
their chronic life stress. It has demonstrated reliability and
validity and has been recommended for use as an outcome
measure of stress .
For the analyses of daily data, we used only full days of
window logging (the time of the study setup sometimes
resulted in partial days of data collection), used weekday
data (i.e. during the work week) and used only days when
the computer usage was greater than zero. For the analyses
of hourly data (investigating the relationship of email
patterns and stress), we used only weekday data, and looked
at average stress (based on RMSSD) and average email use
for each hour during the hours of 9 am to 5 pm, which is
when most participants were in the workplace. We also
used only those hours of data when the computer was used
(i.e. when the logging data showed that computer duration
was greater than zero for that hour).
For our analyses we used Linear Mixed-Effects Models
(LMM) to account for the correlated data within subjects
(repeated measures on days, or on hours). We ran LMM in
SPSS using random and fixed effects. For RQ1, a LMM for
Productivity was based on including independent variables
of Email Duration, Interruption Type, Batching Type, and
control variables. We used a random intercept for
participants; all other factors in the model were entered as
The proportion of seconds spent
daily/hourly on email compared
to total computer duration
Counts of daily/hourly unique
visits to the email client
People's reported preference for
external (use of email
notifications) or self-interruption
for checking email
Based on the daily distribution of
email use, described above
Measured in end-of-day survey
based on six dimensions using
Likert scale; Composite measure
Measured by worn heart rate
monitors using RMSSD
Job demands, job decision
latitude from JCQ , in general
The proportion of seconds spent
daily/hourly on productivity
software compared to total
Perceived Stress Scale  in
Table 1. Summary of measures used.
fixed effects with no random components. For RQ2, a
LMM for HRV (Stress) was based on including the same
independent and control variables as RQ1, with the same
fixed and random effects.
Overview of results with email
The total hours of data collected for window logging was
1981.5, with an average of 49.5 hours of computer screen
data logged per participant. The average number of
weekdays with window activity logged per person (i.e.
excluding Saturdays and Sundays) was 12.4 days.
Table 2 shows that the average daily time spent by our
participants on the computer (averaged over work days) is
about four and a half hours. Our 40 participants averaged
almost one and a half hours per day of time on email and
checked their email on average 77 times per day. 30.8% of
our participants reported primarily checking email due to
external notifications (External), 41.0% reported primarily
checking email on their own (Self), and 28.2% reported
checking email about equally due to external notifications
and on their own. Thirty-four participants had email
notifications enabled. An ANOVA showed no significant
difference in average Email Duration between Self and
External: F(1, 25)=.11, p<.75. Consistents had significantly
longer average Email Duration compared to Batchers and
Mixed strategy: F(2, 37)=6.09, p<.005 and checked email
daily significantly more often: F(2, 37)=6.13, p<.005.
Frequency of Checking Email is highly correlated with
Email Duration: r=.75, p<.0001.
Regression analyses showed no significant relationship of
Job Decision Latitude predicting average Email Duration:
F(1, 38)=2.57, p<.12 but there is a significant negative
correlation with frequency of checking email: F(1,
38)=4.45, p<.04. There is a significant relationship of Job
Demands predicting average Email Duration: F(1,
38)=7.40, p<.01 and a strong positive trend with frequency
of email checking: F(1, 38)=3.58, p<.07. Thus, the higher
employees’ job demands are, the more time they spend on
email and the more often they check email. The more
decision latitude people have in their jobs, the less they
check email. Job Demands and Job Decision Latitude were
controls in our subsequent email analyses.
RQ1. Email use patterns and productivity
We next examine how email is related to information
workers' self-assessed productivity at the end of the day.
Productivity was the dependent variable. Our productivity
index measure (based on 6 dimensions of 7-point Likert
scales) ranged from 6 to 42, M=27.43, SD=7.52.
Independent variables were Email Duration, Interruption
Type, and Batching Behavior. We entered interaction terms
of Interruption Type x Email Duration and Batching
Behavior x Email Duration. Results for RQ1a, RQ1b and
RQ1c are shown in Tables 3a and 3b. Control variables of
Job Demands, Job Decision Latitude, and Productivity
software duration were not significant. Between subjects
variance was 21.79, SE=2.76, and within-subjects variance
was 23.03, SE=9.07.
RQ1a. Email Duration
Email Duration is significantly negatively related to
Productivity: the more time spent on email for that day, the
lower the assessed productivity for that day (Table 3b).
RQ1b. Interruption Type
There were no main effects of Interruption Type. However,
Interrupt Type x Email Dur
Batching Type x Email Dur
Table 3a. Model of Email use patterns with Productivity:
tests of fixed effects of Email Duration, Interruption Type
and Batching Behavior.
RQ1a: Email duration
RQ1b: Interruption Type: Self 1
RQ1c: Batching Type: Consistent2
RQ1c: Batching Type: Batching2
Self Interruption x Email Duration1
2.64 (.95) **
Consistent x Email Duration2
Batching x Email Duration2
3.33 (1.35) *
Table 3b. Coefficients and SE of fixed effects in Table 3a.
***p<.001, **p<.01, *p<.05. 1External-interruptions is the
reference category; 2Mixed strategy is the reference
4 hr 34
2 hr 23
4 hr 28
3 min - 13
hr 59 min
1 hr 23
1 hr 6
0 - 7 hr 54
1 - 408
Table 2. Daily averages of different computer usage. N=40.
there was a significant Interruption Type x Email Duration
interaction. As Email Duration increases, productivity is
rated highest by those who check email on their own
compared to those who primarily rely on email notifications
RQ1c. Email Batching behavior
Batching behavior as a main effect was not significantly
related to Productivity. However, there was a significant
interaction of Batching behavior and Email Duration.
Batchers rate their productivity higher at high email
durations in relation to a mixed strategy.
Explaining email duration and productivity
Even though the main effect of Email Duration shows a
negative relationship with Productivity, the interaction
results show interesting patterns of Self-interrupters and
Batchers rating their productivity higher with more time
spent on email compared to the reference groups. The
results of Email Duration could be due to some people
considering themselves to be more productive when using
email than others. Even though we normalized by person
and controlled for job characteristics and productivity
software, it is still possible that some workers, more than
others, may view their time on email as accomplishing
work and therefore feel more productive the longer their
email use. To check this notion, we compared the ten
participants with the highest average daily productivity
ratings (averaged over all days in the study) with the ten
participants with the lowest average daily productivity
ratings, to see if email duration differed. An independent t-
test showed that there was no significant difference between
the two groups, t(18)=.26, p<.80). Similarly, there was no
significant difference between the two groups in average
daily Email Checks: t(18)=.93, p<.37. Therefore, though
some people rate their productivity higher than others, the
relationship of email duration and productivity rather varies
within individuals, i.e., when a person spends more time on
email relative to their mean usage, then their productivity
assessment declines (and vice versa).
RQ2. Email use patterns and stress
We next examined the relationship of email usage patterns
and stress. As described in the methods section, stress is
measured by HRV, based on the heart rate captured by the
worn heart rate monitors. Recall that the value of HRV is
inversely related to one's stress level: the lower the HRV
value, the higher the stress. Email duration was compared
with average HRV (using the measure of RMSSD) for that
same hour, during the work hours of 9 to 5. Results are
shown in Table 4a and 4b. For all analyses, we controlled
for Job Demands, Job Decision Latitude, and baseline
stress, as measured by the Perceived Stress Scale (PSS)
instrument . Control variables were not significant.
Between subjects variance was 83.09, SE=4.01, and within-
subjects variance was 129.10, SE=43.64.
RQ2a. Email Duration
Email Duration is significantly associated with Stress. The
longer one spends on email that hour, the higher is one's
stress for that hour (Table 4b).
RQ2b. Interruption Type
Interruption Type did not show a significant relationship
with Stress as a main effect nor was there a significant
interaction with Email Duration.
RQ2c. Email batching behavior
Batching behavior was neither significant as a main effect
on Stress, nor was there a significant interaction of
Batching behavior and Email Duration on Stress.
There has been increasing interest in HCI on the role of
interruptions in work, both externally triggered and self-
initiated. However, to our knowledge, no one has explored
the relationship of email usage patterns of duration of time
on email, interruption habits (notifications or self-
interruptions), and batching behavior, and how they relate
to perceived workplace productivity or stress.
While studies of email use generally involve self-reports,
we used an approach with more sophisticated measures,
where email usage was logged in situ in the workplace, and
participants' internal states were captured via physiological
Interrupt Type x Email Dur
Batching Type x Email Dur
Table 4a. Model of Email use patterns with HRV: tests of
fixed effects of Email Duration, Interruption Type and
RQ2a: Email duration
RQ2b: Interruption Type: Self 1
RQ2c: Batching Type: Consistent2
RQ2c: Batching Type: Batching2
Self Interruption x Email Duration1
Consistent x Email Duration2
Batching x Email Duration2
Table 4b. Coefficients and SE of fixed effects in Table 4a.
**p<.01. 1External-interruptions is the reference category;
2Mixed strategy is the reference category. Note that the
lower the value of HRV, the higher the stress.
measures and self-reports. This enabled us to not only
examine email usage based on objective logged data, but to
also complement our analysis with measures of participants'
cognitive and affective states.
Email duration showed significant effects both with
productivity and stress. When an individual spends more
time on email during the workday, it is significantly related
to lower assessed productivity and higher stress, after
controlling for job characteristics. Our findings of email
duration build on results of Hanrahan and Pérez-Quiñones
, in that the more time on email, the more opportunities
there are for diversions within the email client. Iqbal and
Horvitz  found that it took over nine minutes to return
to an interrupted task from checking email, when diversions
extended beyond the email client. More diversions from
one's task-at-hand could lead one to feel less productive and
Though we did not study cognitive overload per se, our
results with Email Duration and stress also build on other
studies of email use that find that email use is associated
with a feeling of cognitive overload: due to poor email
management strategies [9, 48], coordination challenges that
email introduces , the work that email invites , and
social pressures to respond [2, 41].
We found that different email use patterns interact with
duration of time on email and were associated with
perceived productivity and stress. Participants who
primarily checked email of their own volition reported
higher productivity at days' end with higher email duration,
compared to those who primarily check email through
notifications. One reason could be that people who
primarily self-interrupt to do email could feel that they have
more agency in their work, i.e., by choosing when to
interrupt, they could feel more productive. Also,
interruptions involve a significant recovery time to reorient
back to an interrupted task [18, 35]. Perhaps people who
self-interrupt have more control over when to take a task
break, e.g. gearing their interruptions to natural break points
in tasks  making it easier to resume an interrupted task,
thus leading to a higher feeling of productivity. Those who
check email on their own may be better at adapting their
use of email based on the state of their ongoing tasks.
Many claims in the popular media tout that batching email
should lead people to be more productive and feel less
stress. Our study found some support for these claims for
productivity but only with high email use. One explanation
could be that there are diminishing marginal returns for
checking consistently. Perhaps as time spent on email
increases, consistently checking ends up wasting time,
providing less relevant information relative to the time
invested. At high email durations, batching may yield better
returns and may be perceived as more efficient. This
relationship warrants further examination. However,
contrary to claims that batching email can reduce stress, we
found no evidence in our study to suggest this.
Personality: exploring email usage patterns
Our results raise the question of why batchers and self-
interrupters feel more productive at high email volume.
Although it was not one of our primary research questions,
we opted to explore this post facto, and wondered whether
we might be able to explain users’ choice of email
strategies based on individual differences. As a first step,
we explored personality, since we had collected scores on
the Big 5 personality traits  in the general survey.
Specifically, we expected that Consistents (non-Batchers)
and Self-interrupters might score higher on the
Conscientiousness trait of the Big 5 personality test , as
it describes people who are careful and vigilant. We did not
find differences with Self and External Interruption Types.
However, in comparing Batchers with Consistents, we
found that Consistents score significantly higher in the
Conscientiousness personality trait: (F(1, 32)=12.97,
p<.001. Using logistic regression, the Nagelkerke R2
showed that Conscientiousness explains 36.8% of the
variance of Batching/non-Batching behavior. This
exploratory result suggests that different email management
strategies could be related to personality differences, worth
Implications for organizations
Our study is unique in that we found a relationship with
increasing time that people spend on workday email and
higher stress. Cutting off email in the workplace has been
found to lower stress  as was limiting the frequency of
checking email . Neither of these conditions is realistic
for the workplace given social norms . Our study instead
examined in situ, naturalistic workplace behavior. Until we
invent a better replacement, email will not go away.
While email use certainly saves people time and effort in
communicating, it also comes at a cost. Our results suggest
implications for organizations: spending longer time on
email may have detrimental effects in the form of
workplace stress . Any intervention that can decrease
stress is beneficial. Future research could examine more
carefully exactly what types of workplace activities might
be traded off with email use. Of course many factors of
email can influence productivity and stress in the
workplace. For example, receiving timely and relevant
information, the job role of the sender, and the tone of the
received email can all influence productivity and stress.
Our study is a first step in providing evidence that suggests
that reducing time on email could be beneficial.
Our findings can benefit organizations. Cutting down on
email time (associated with higher assessed productivity
and less stress) could improve the health and the wellbeing
of employees. First, we suggest that organizations make a
concerted effort to cut down on email traffic. Organizations
could use a pull channel or wikis for much organizational
information, reducing the volume of email. Second, while
batching does not offer benefits for short durations on
email, it may be a good strategy for those who expect to
receive a large volume of email, as it will result in fewer
interruptions (volume is shown to be correlated with
duration ). Third, while self-discipline can be a
challenge, perhaps if employees are made aware that time
on email can lead to stress, this could motivate them to
restrict email time. Tools and user interface designs for
protecting stressed users from the onslaught of email could
contribute to improving workplace productivity and health.
Notes on causality
We found that patterns of email use is related to some
aspects of the workplace experience, yet our data is
correlational, which does not imply causality. One way to
support causality is to find converging evidence for a
phenomenon. Our results on stress are consistent with other
studies that have shown a positive relationship between
email and stress, e.g., [2, 28]. We have additionally shown
that the duration of time spent on email is associated with
stress. Of course, it is also possible that the causality works
in the opposite direction. For example, people may first
assess themselves as unproductive (i.e., at the beginning of
the day), and as a result may then engage in more time on
email. We find that this argument is not convincing. First,
people assessed their productivity at the end of the day, and
we assume that they were considering an overview of how
productive they felt throughout the day, assessed at that end
of the day moment. There is thus a time relationship of
email duration measured throughout the day along with a
productivity assessment at the end of the day. Second,
numerous studies have documented the varied activities that
email involves that can lead to extra peripheral work, e.g.
[2, 3, 10, 47, 48]. However, controlled experiments would
be needed to disentangle the causality.
As we move into an era of Big Data analytics (we consider
our data using computer logging and stress tracking "small"
Big Data), varied questions of correlation and causation
arise. As correlation does not imply causality we cannot
ascertain the direction of the relationships that we found.
Whereas laboratory studies enable the manipulation of
variables to assess causation, in situ tracking captures real
world IT usage from multiple perspectives. We feel the two
methods are complementary: tracking studies can identify
phenomena that can be then tested in the laboratory.
Our study has several limitations. We looked at the time
duration of email and the checking patterns but did not look
at the content of the email, as our field site did not allow
this. For example, email that assigns tasks or that is from
one's superior might lead to higher stress than other types of
email, e.g. personal email. Thus, we cannot make
inferences into how email content might affect overload and
the workplace experience. This remains as future work but
obviously could have privacy implications.
We deliberately bounded our study to email use in the
workplace. We did not examine email usage outside of the
workplace hours, and individuals could use time outside of
the workplace to manage emails that they could not get to
during the day . This is again a topic for future research.
Despite the fact that we made every effort to gain an
accurate measure of email use through logging, we cannot
capture email use 100%. If people look away from their
email, the logger does not capture this. However, mouse or
keystroke activity did serve as a check that email was being
used, so we are reasonably confident that we have a good
representation of email use. Further, objective logging of
email is far more accurate than self-reports, which many
studies rely on, cf . Also, some of our participants used
phones to read email and our Windows Activity logger did
not work on phones. However, all participants reported to
us that their primary way of accessing email was on their
laptops or desktops, which we logged.
Our participants were from a single workplace. Although
they were in a variety of job roles and their job
characteristics expanded across a wide range, we must be
careful when generalizing this across other workplaces.
Professional context could also play a role in email use .
Our results apply to large organizations involving
information work. The information workers in our study
were highly educated, having at least a bachelor's degree, so
we can only generalize the results to similar people.
Why then do people spend time on email if it is associated
with feeling less productive and more stressed? Numerous
studies have highlighted the benefits of email. There are
social reasons , the need to keep on top of email to get
critical information , there are social norms to respond
(quickly) [2, 41], power dynamics in the workplace, and a
host of other reasons. Thus, in our current workplace
environment, we need email, but it comes at a cost.
Email is clearly an important part of the work life of
information workers. An accumulating body of empirical
research as well as anecdotal evidence shows that the
benefits of email use come at a cost, however, of impacting
users’ wellbeing. Our study contributes to this body of
research by focusing on the relationship between email use
and two key variables important to the workplace
experience: productivity and stress. Our results benefited
from capturing email usage from both external (logging and
physiological) measures and internal user perspectives,
which enabled us to investigate fairly nuanced in situ
experiences. As the development of measurement
techniques continues to expand, we expect them to reap
deeper understandings of people's in situ workplace
experiences. We hope that our study can spark future
research directions for email management systems that can
benefit work with less cost to the user, and that can improve
This material is based upon work supported by the NSF
under grant #1218705.
1. U. Rajendra Acharya, K. Paul Joseph, N. Kannathal, Choo
Min Lim, and Jasjit S. Suri, 2006. Heart rate variability: A
review. Medical & Bio Engineering & Comp 44, 12,
2. Stephen R. Barley, Debra E. Meyerson, and Stine Grodal.
2011. E-mail as a source and symbol of stress.
Organization Science 22, 4, 887–906.
3. Victoria Bellotti, Nicolas Ducheneaut, Mark Howard, Ian
Smith, and Rebecca E. Grinter, R.E. 2005. Quality versus
quantity: E-mail-centric task management and its relation
with overload. Human-Computer Interaction 20, 1, 89–
4. Adam Bradley, Duncan P. Brumby, Anna L. Cox, and Jon
Bird. 2013. How to manage your inbox: is a once a day
strategy best? Proceedings of the 27th International BCS
Human Computer Interaction Conference, British
Computer Society, 20-26.
5. Marta E. Cecchinato, Anna L. Cox, and Jon Bird. 2015.
Working 9-5?: Professional Differences in Email and
Boundary Management Practices. Proceedings of the 33rd
Annual ACM Conference on Human Factors in
Computing Systems (CHI '15). ACM, New York, NY,
USA, 3989-3998. DOI=10.1145/2702123.2702537
6. Sheldon Cohen, Tom Kamarck, and Robin Mermelstein.
1983. A global measure of perceived stress. Journal of
Health and Social Behavior, 24, 385-396
7. Fred Collopy. 1996. Biases in retrospective self-reports of
time use: An empirical study of computer users.
Management Science. 42, 758-767.
8. Mary Czerwinski, Eric Horvitz, and Susan Wilhite. 2004.
A diary study of task switching and interruptions.
Proceedings of the SIGCHI Conference on Human
Factors in Computing Systems(CHI '04). ACM, New
York, NY, USA, 175-182. DOI=10.1145/985692.985715
9. Laura A. Dabbish and Robert E. Kraut. 2006. Email
overload at work: an analysis of factors associated with
email strain. Proceedings of the 2006 20th anniversary
conference on Computer supported cooperative
work (CSCW '06). ACM, New York, NY, USA, 431-440.
10. Nicolas Ducheneaut and Victoria Bellotti. 2001. Email as
habitat: An exploration of embedded personal information
management. ACM Interactions. September-October; 8
11. Nicolas Ducheneut and Leon A.Watts. 2011. In search of
coherence: A review of email research. Human-Computer
Interaction, 20:1-2, 11-48.
12. Angela Edmunds and Anne Morris. 2000. The problem of
information overload in business organizations: a review
of the literature. International journal of information
management, 20(1), 17-28.
13. Martin J. Eppler, and Jeanne Mengis. 2004. The concept
of information overload: A review of literature from
organization science, accounting, marketing, MIS, and
related disciplines. The information society, 20(5), 325-
14. Danyel Fisher, A. J. Brush, Eric Gleave, and Marc A.
Smith. 2006. Revisiting Whittaker & Sidner's "email
overload" ten years later. Proceedings of the 2006 20th
anniversary conference on Computer supported
cooperative work (CSCW '06). ACM, New York, NY,
USA, 309-312. DOI=10.1145/1180875.1180922
15. Mario Hair, Karen Renuad, and Judith Ramsay. 2007. The
influence of self-esteem and locus of control on perceived
email-related stress. Computers in Human Behavior, 23
(6), November 2007, 2791–2803.
16. Benjamin V. Hanrahan and Manuel A. Pérez-Quiñones.
2015. Lost in Email: Pulling Users Down a Path of
Interaction. Proceedings of the 33rd Annual ACM
Conference on Human Factors in Computing
Systems (CHI '15). ACM, New York, NY, USA, 3981-
17. Shamsi T. Iqbal and Eric Horvitz. 2010. Notifications and
awareness: a field study of alert usage and preferences.
Proceedings of the 2010 ACM conference on Computer
supported cooperative work (CSCW '10). ACM, New
York, NY, USA, 27-30. DOI=10.1145/1718918.1718926
18. Shamsi T. Iqbal and Eric Horvitz. 2007. Disruption and
recovery of computing tasks: field study, analysis, and
directions. Proceedings of the SIGCHI Conference on
Human Factors in Computing Systems (CHI '07). ACM,
New York, NY, USA, 677-686.
19. Thomas W. Jackson, Ray Dawson, and Darren Wilson.
2003. Understanding email interaction increases
organizational productivity. Communications of the
ACM 46, 8 (August 2003), 80-84.
20. Jing Jin and Laura A. Dabbish. 2009. Self-interruption on
the computer: a typology of discretionary task
interleaving. Proceedings of the SIGCHI Conference on
Human Factors in Computing Systems (CHI '09). ACM,
New York, NY, USA, 1799-1808.
21. Robert Karasek, Chantal Brisson, Norito Kawakami,
Irene Houtman, Paulien Bongers and Benjamin Amick.
1998. The Job Content Questionnaire (JCQ): an
instrument for internationally comparative assessments of
psychosocial job characteristics. Journal of occupational
health psychology, 3(4), 322-355. DOI= 10.1037/1076-
22. Pamela Karr-Wisniewski and Ying Lu. 2010. When more
is too much: Operationalizing technology overload and
exploring its impact on knowledge worker productivity.
Computers in Human Behavior, 26, 1061-1072.
23. Kostadin Kushlev and Elizabeth W. Dunn. 2015.
Checking email less frequently reduces stress. Computers
in Human Behavior, 43, 220-228.
24. Daniel J. Levitin, The organized mind: Thinking straight
in the age of information overload. Penguin, 2014.
25. Rita Mano and Gustavo Mesch. 2010. Email
characteristics, work performance and distress. Computers
in Human Behavior, 26, 61-69.
26. Gloria Mark, Shamsi T. Iqbal, Mary Czerwinski, and Paul
Johns. 2015. Focused, aroused, but so distractible: A
temporal perspective on multitasking and
communications. Proceedings of CSCW 2015, ACM
27. Gloria Mark, Shamsi T. Iqbal, Mary Czerwinski, and Paul
Johns. 2014. Bored Tuesdays and focused afternoons: The
rhythm of attention and online activity in the workplace.
Proceedings of CHI’14, ACM Press, 3025-3034.
28. Gloria Mark, Steven Voida, and Armand Cardello. 2012.
“A Pace Not Dictated by Electrons”: An Empirical Study
of Work Without Email. Proceeding of CHI’12, ACM
29. Robert McCrae and Paul Costa. 1999. The five factor
theory of personality. in Handbook of Personality: Theory
and Research, L.A. Pervin, O.P. Johns, NY: Guilford,
30. Laura Marulanda-Carter, and Thomas Jackson. 2012.
Effects of e-mail addiction and interruptions on
employees. J of Systems and Information
31. Laura Marulanda-Carter, Thomas Jackson, G. Ragsdell.
2010. Developing an email stress measuring methodology
to determine the impact email stress has on employee
effectiveness. Proceedings of the SQM Conference, March
32. Melissa Mazmanian. 2013. Avoiding the trap of constant
connectivity: When congruent frames allow for
heterogeneous practices. Academy of Management
Journal 56.5 (2013): 1225-1250.
33. McKinsey (2012). http://www.mckinsey.com/insights-
34. Catherine Middleton and Wendy Cukier. 2006. Is mobile
email functional or dysfunctional? Two perspectives on
mobile email usage. European Journal of Information
Systems, 15(3) 252–260.
35. Karen Renaud, Judith Ramsay, and Mario Hair. 2006.
"You've got e-mail!"... shall I deal with it now? Electronic
mail from the recipient's perspective. International
Journal of Human-Computer Interaction, 21(3), 313-332.
36. Stever Robbins. 2004.Tips for mastering e-mail overload.
Harvard Business School Working Knowledge (2004, Oct
37. Gokturk Sevinc and John D'Ambra. 2010. The Influence
of Self-Esteem and Locus of Control on Perceived Email
Overload. Proceedings of ECIS 2010.
38. Cheri Speier, Joseph S. Valacich, and Iris Vessey. 1999.
The influence of task interruption on individual decision
making: An information overload perspective. Decision
Sciences, 30(2), 337-360.
39. David Sumecki, Maxwell Chipulu, and Udechukwu
Ojiako. 2011. Email overload: Exploring the moderating
role of the perception of email as a ‘business critical
tool. International Journal of Information
Management 31.5 (2011): 407-414.
40. Agnieszka Matysiak Szóstek. 2011. 'Dealing with My
Emails': Latent user needs in email
management.. Comput. Hum. Behav. 27, 2 (March 2011),
41. Gail Fann Thomas, Cynthia L. King, Brian Baroni, Linda
Cook, Marian Keitelman, Steve Miller, and Adelia
Wardle. 2006. Reconceptualizing e-mail overload. J of
Business and Technical Communication, 20(3), 252-287.
42. Sara Thomée, Annika Härenstam, and Mats Hagberg.
2012. Computer use and stress, sleep disturbances, and
symptoms of depression among young adults–a
prospective cohort study. BMC psychiatry 12.1 (2012):
43. Ian Towers, Linda Duxbury, Christopher Higgins, and
John Thomas. 2006. Time thieves and space invaders:
technology, work and the organization. Journal of
Organizational Change Management, 19(5), 593-618.
44. Sherry Turkle. 2010. Alone Together. MIT Press,
45. Tanja Vrijkotte, , Lorenz Van Doornen, and Eco De Geus.
2000. Effects of work stress on ambulatory blood
pressure, heart rate, and heart rate variability.
Hypertension, 35(4), 880-886.
46. David Watson, Lee A. Clark, and Auke Tellegen. 1988.
A. Development and validation of brief measures of
positive and negative affect: The PANAS scales. J of
Personality and Social Psychology, 56 (6). 1063-1070.
47. Steve Whittaker, Victoria Bellotti, and Jacek Gwizdka.
.2006. Email in personal information management.
Communications of the ACM, 49(1), Jan. 2006, 68-73.
48. Steve Whittaker and Candace Sidner. 1996. Email
overload: Exploring personal information management of
email. In Proc. CHI 1996, ACM Press, 276–283.