IEEE Technology and Society Magazine - Winter 2014 - 67

[13], which took place in 2005.
However, earlier case-crossover
studies compared cell phone use
during the case window, or the
time before a crash, and also during
the control window, the same time
period on a previous day. A later
study controlled for those who may
have been driving for only part of
the control window by collecting
actual GPS data from instrumented
cars [31]. That later study found
that the final adjusted relative risk
was lower (compared to earlier
studies) at 1.1.
Taking a longitudinal approach,
another recent study [1] used two
survey data sets, one from 1997
and one from 2007, to assess the
relative risk of cell phone use
while driving. The data conveyed a
higher and more significant relative
risk (2.34) for both hands-free and
handheld cell phones combined.
However, the 2007 data showed
higher yet less significant relative
risk estimates. The authors suggested a "learning effect" may be
responsible for the potential drop
in crash risk from 1997 to 2007 [1].
Further research was conducted
through an approach known as
meta-study. For example, a metaanalysis of 12 studies determined
that, out of the studies that have
reliable information about cell
phone use at the time of a crash
(e.g., cell phone company billing
records), the summary estimate of
the relative risk is 2.86 [7], but it
is possible that the studies did not
adequately control "all potentially
important confounding factors"
[p. 25] and that these studies overestimated the crash risk.
Epidemiological studies have
generally revealed an increased
crash risk associated with cell
phone use. However, some results,
such as the insignificant risk estimates using the 2007 data [1], do
not lend themselves to definitive
conclusions. It is possible that the
risk estimates produced by the studies included effects of confounding
variables that were not controlled

[7]. In response, there have been
calls for the integration of more
relevant variables into the analysis
[3]. While epidemiological studies
do provide insight into the association between these variables, they
do not definitively establish causality between cell phone use and
accident rate.
Observational Studies
Observational studies represent a
tradeoff between external validity (e.g., actual roads) and internal validity (e.g., measurements),

IEEE TECHNOLOGY AND SOCIETY MAGAZINE

and offer another way to increase
our understanding of cell phone
use and crash risk. Observational
studies are scientific investigations
that take place in actual settings.
One type of observational study
uses the naturalistic observation of
drivers in the real world. Typically,
volunteers are recruited and their
personal cars wired with cameras
recording their activities over a
period of time. While this method
cannot technically "manipulate"
an independent variable (e.g., telling the driver to use a cell phone),

Table I
Summary of Literature Reviewed
Authors

Findings

Drews
et al. [5]
Dula
et al. [6]

Strayer
et al. [29]

McEvoy
et al. [13]

Young [31]

BackerGrøndahl
and
Sagberg [1]

Elvik [7]

Klauer
et al. [9]
Klauer
et al. [10]

WINTER 2014

Conversations on cell phones
affect driving performance more
than those with in-car passengers.
More emotional conversations
on cell phone produces more
dangerous driving behaviors in
four of five driving variables.
Using speech-to-text technology
was the most cognitively
distracting when compared to other
in-car activities.
Cell phone use while driving
increases crash risk by a factor
of 4. Risk increased significantly
regardless of handheld or
hands-free device.
The final adjusted relative risk was
1.1 (lower than earlier studies) after
controlling for those who may have
been driving for only part of the
control window.
The 1997 data showed higher,
significant relative risk for both
hands-free and handheld cell
phones combined (2.34), but the
2007 data showed higher but nonsignificant relative risk estimates.
The summary estimate of the relative
risk is 2.86 based on studies that have
reliable information about cell phone
use at the time of a crash (e.g., cell
phone company billing records),
but it is possible that the studies
overestimated the crash risk.
Dialing a handheld device
increased risk of crash or
near-crash by 2.8 times.
For novice drivers, dialing a
handheld device increased risk of
crash or near-crash by 8.3 times.
For experienced drivers, dialing
a handheld device increased the
same risk by 2.5 times.

Type of study
Experimental

Experimental

Epidemiological

Epidemiological
(meta-analysis)

Observational

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - Winter 2014