You are on page 1of 7

Forensic Science International: Digital Investigation 37 (2021) 301167

Contents lists available at ScienceDirect

Forensic Science International: Digital Investigation


journal homepage: www.elsevier.com/locate/fsidi

Exploring digital evidence recognition among front-line law


enforcement officers at fatal crash scenes
Thomas Holt a, *, Diana S. Dolliver b
a
Criminal Justice Department, Michigan State University, USA
b
The Department of Criminology and Criminal Justice, The University of Alabama, USA

a r t i c l e i n f o a b s t r a c t

Article history: Digital forensic evidence has become an essential component of criminal investigations across the United
Received 9 July 2020 States. The amplified integration of technology with society, however, continues to present new chal-
Received in revised form lenges for law enforcement officers. This issue is evident in one of the key responsibilities of police
29 March 2021
agencies involves the enforcement of traffic laws and regulations on surface streets and highways. It is
Accepted 23 April 2021
Available online 7 May 2021
unclear how well first responders recognize the need to seize digital evidence associated with fatal
vehicles accidents in the field. This study addressed this gap in the literature by examining a comparative
population of respondents from the state police agencies in a large Midwestern and large Southern state.
Keywords:
Crime scene
Using a vignette design, the findings demonstrated that respondents recognized the need to recover
Digital evidence mobile devices and telematics from the vehicles involved in fatal crashes. A substantive number of of-
Law enforcement ficers remained unsure of the potential digital evidentiary value of other objects on-scene (e.g., fast food,
State police makeup). Further, regional differences were observed in respondents’ responses. The implications for our
Technology understanding of police capacity and training needs were explored in detail.
© 2021 Elsevier Ltd. All rights reserved.

1. Introduction (National Highway Transportation Safety Board, 2019; Rivers,


2011). In recent years, there has been concern raised about the
One of the key public order roles of police agencies involves the potential for accidents stemming from drivers who are distracted
enforcement of traffic laws and regulations on surface streets and by their digital devices, including cell phones, MP3 players, and GPS
highways (DeAngelo and Hansen, 2014; Van Kirk, 2000; Walker technology (Chan and Singhal, 2015; Overton et al., 2015; Wilson
and Katz, 2012). Police are often called in the event of traffic acci- and Stimpson, 2010). Research from the National Highway Traffic
dents, such as vehicle collisions, to assist injured parties, determine Safety Administration estimates that approximately two percent of
if laws were broken, and help keep order at the scene. This role is the driving population in the US engages in hand-held manipula-
particularly vital in the event of serious injuries to drivers or pas- tion of mobile devices while driving at any time (National Center
sengers, or fatalities that may have resulted from any collision or for Statistics and Analysis, 2019). Additionally, the rate of manip-
crash (DeAngelo and Hansen, 2014; Walker and Katz, 2012). Offi- ulation of devices has increased across all age groups since 2006,
cers collect information from drivers, passengers, and witnesses at particularly among those aged 16 to 24 (National Center for
the scene to help reconstruct the events leading up to the accident Statistics and Analysis, 2019). Due to the risks posed by distracted
(Rivers, 2011; Van Kirk, 2000). The investigation can also determine driving, 47 states and the District of Columbia have banned text
what laws were violated, as well as who may be charged for what messaging for drivers backed by traffic citations that can be issued
offenses (Rivers, 2011; Van Kirk, 2000). by officers (National Center for Statistics and Analysis, 2019).
Thousands of fatal crashes occur in the US ever year, stemming The rise of so-called connected or autonomous vehicles have
from a range of factors including the failure of critical vehicle parts also introduced potential for accidents, as these vehicles can be
to inclement weather affecting driving conditions to driver error guided by microchips and electronics without the need for constant
driver attention (Dimitrakopoulos, 2011; Gerla et al., 2014;
Kennedy et al., 2019). Auto manufacturers such as Tesla argue that
* Corresponding author. drivers should be aware of their surroundings and able to take
E-mail addresses: holtt@msu.edu (T. Holt), DLDolliver@protonmail.com control of the vehicle at all times, though evidence suggests that is
(D.S. Dolliver).

https://doi.org/10.1016/j.fsidi.2021.301167
2666-2817/© 2021 Elsevier Ltd. All rights reserved.
T. Holt and D.S. Dolliver Forensic Science International: Digital Investigation 37 (2021) 301167

not the case (Kennedy et al., 2019; Schmelzer, 2019). In fact, several portable storage devices (i.e., compact disc, external hard drive,
fatal crashes have occurred where an automated vehicle was in use flash drive), wearable technologies (e.g., smart watches), and smart
and the driver was distracted or otherwise unable to avoid the televisions (Goodison et al., 2015; Novak et al., 2018). These items
accident (Kennedy et al., 2019; Schmelzer, 2019). hold tangible evidence of criminal acts, such as documents, files,
As technology increasingly affects both the operation and op- and images that may implicate an individual in an act (Casey, 2011;
erators of automobiles, it is essential that police be able to use the Dolliver et al., 2017; Tuner, 2019). Additionally, digital evidence
information stored on these devices when investigating fatal provides supply routes to information that perpetrators may have
crashes (Kennedy et al., 2019; Rivers, 2011). Specifically, the mobile tried to destroy via cache review and file recovery tools (Lupariello
devices available to the driver and passengers in vehicles are et al., 2018; Tene, 2008). There is also a wealth of information
essential to investigators, including the text messages, web search gathered from digital technologies citizens interact with every day,
histories, applications, photos, and GPS data, would be essential for such as automatic license plate readers, social media search tools,
accident investigators (Casey, 2011; Goodison et al., 2015). The cell site location data, GPS tracking, and surveillance cameras
devices and equipment used to transmit information can also serve located in public and private spaces (Ferguson, 2017; Tuner, 2019).
as sources of digital evidence, such as wireless routers and web Digital evidence plays a particularly important role at the scenes
servers the devices may have interacted with, including any in- of fatal vehicle accidents, which are a key investigative re-
vehicle wireless Internet hotspots (Kennedy et al., 2019; Maras, sponsibility for state police, namely highway patrol officers and
2012). The onboard computer and infotainment systems within troopers (Rivers, 2011; Van Kirk, 2000). In 2018, approximately
vehicles also produce digital evidence that can be used to support 36,750 individuals lost their lives in vehicle accidents in the United
system failures and driver behaviors (Gerla et al., 2014). States, at a rate of roughly 1.14 deaths per 100 million vehicle miles
The potential value of digital evidence in criminal justice prac- traveled (National Highway Traffic Safety Administration, 2019). To
tice has led police and policy organizations to create guidelines and determine the cause of such accidents, responding officers must
recommendations for best practices in the field to ensure it is take the totality of circumstances into account, including deter-
correctly seized and stored (IACP, 2019; NIJ, 2008). These materials mining the extent of distractions occurring within the vehicle(s)
are targeted toward the investigation of traditional criminal of- involved, weather and road conditions, and the vehicle's actions
fenses, and research suggests line officers in police agencies are not leading up to the crash (Rivers, 2011; Van Kirk, 2000). However, it is
well prepared to identify or secure digital evidence in the field unclear how well troopers in the United States are able to correctly
(Dolliver et al., 2017; Goodison et al., 2015; Holt et al., 2010, 2015, identify and properly seize digital evidence from these types of
2019; Stambaugh et al., 2001). There is even less knowledge as to crime scenes.
how digital evidence should be used in the investigation of fatal Since approximately one-quarter of traffic accidents involve the
vehicle crashes, or how well first responders recognize the need to usage of cell phones by the driver to engage in (often illegal) ac-
seize digital evidence associated with vehicles in the field (Dolliver tivities, such as “texting and driving” (Chan and Singhal, 2015;
et al., 2017; Holt et al., 2019). Wilson and Stimpson, 2010), it is essential to forensically examine
This exploratory study addressed this gap in the literature the mobile devices of those involved in fatal vehicle accidents to
through an examination of a population of line troopers and agents establish cause (Horsman and Conniss, 2015). Data from mobile
who respond to fatal accidents from two state police agencies. device connected to a vehicle may yield digital evidence in the form
Respondents were presented with a scenario requiring them to of SMS and MMS messaging, call logs, email applications, Internet
identify what digital and physical evidence they would collect at browsing history, and social media application usage (e.g., Snap-
the scene of a fatal accident. The experiential and demographic chat, Twitter, Instagram, Facebook). Active and passive interaction
factors associated with the need to secure digital evidence were with a mobile device while driving is indicative of boredom and
examined, demonstrating that respondents with prior training and lapses in concentration, frequent causes of vehicle accidents
digital evidence field experience were more likely to seize digital (Oulasvirta et al., 2012; Overton et al., 2015). As such, these data
evidence. The implications of this analysis for our understanding of may be used by investigators to compose a timeline of actions taken
police practice and policy are explored in detail. by the driver leading up to the moment of the fatal crash.
Less apparent forms of digital forensic evidence pertinent to
2. Digital evidence: the benefits and challenges in law crash investigations are the data stored on the vehicle's electronic
enforcement control units (ECUs). ECUs are embedded specialized computers
that control all of the electronic functions of a car; there are
In the United States, technology is an essential part of social life. approximately 70 ECUs and 5 to 7 separate in-vehicle networks per
Most U.S. adults own cell phones (96%), smartphones (81%), com- automobile (Nilsson et al., 2018). ECUs of particular importance to
puters (75%), and e-readers (~50%) (Anderson, 2019). The varied use investigators are the vehicle's on-board infotainment and tele-
of technology and its integration into our personal and professional matics systems. Infotainment units are a combination of hardware
lives creates an environment where individuals are overtly doc- and software that provide in-car audio and video entertainment as
umenting their everyday lives through social media platforms and toggled by the vehicle's occupants (i.e., data that stays within the
blogging (Anderson and Jiang, 2018; Dolliver, 2019; Holt and vehicle), while telematics systems are the telecommunication and
Bossler, 2015). The devices and platforms used for communica- information hub of the vehicle (Kennedy et al., 2019; Nilsson et al.,
tion also covertly and overtly track users’ movements in space and 2018). The telematics ECU is also essential as a source of informa-
time (Dolliver, 2019; Novak et al., 2018). The search terms, photos, tion as it connects the vehicle to external networks, such as cellular,
texts, and emails sent by individuals also create social artifacts that Bluetooth, and WiFi. At the same time, the telematics ECU connects
indicate their behaviors and interests (Tene, 2008). the vehicle to infrastructure (called V2I), such as interactions with
As a result, digital evidence has become an increasingly smart traffic lights, as well as vehicle-to-vehicle (V2V) communi-
important source of information in proactive and reactive policing cations, as observed in autonomous and self-parking vehicles
encounters. The term “digital evidence” is intentionally ambiguous (Nilsson et al., 2018).
given the breadth of information investigators can glean from Data stored on both the telematics and infotainment ECUs are
anything that connects to an electronic network and/or stores relevant to crash investigations. For instance, they maintain vehicle
electronic information: computer, cellular phone, digital camera, system information such as the VIN and serial numbers of vehicle
2
T. Holt and D.S. Dolliver Forensic Science International: Digital Investigation 37 (2021) 301167

systems. They also capture vehicle events such as the speed of the instance, 311 respondents indicated their rank or pay level, which
vehicle, airbag deployment, doors opening and closing, GPS reflects a response rate of 17.7% of all active sworn forces (n ¼ 1751).
syncing, breaking, and peripheral device connections (Kennedy Additionally, 51.1% of all respondents served as troopers, which is in
et al., 2019; Nilsson et al., 2018). The information of any mobile keeping with the general distribution of troopers in the overall
device connected or paired to the vehicle, whether via USB cable or force (n ¼ 1022; 58%). Respondents were predominantly male
Bluetooth, is also captured, including device IDs, call logs, SMS (n ¼ 275; 88.4%), and white (n ¼ 280; 81.2%), with African Ameri-
messages, and navigation data (e.g., track-logs, saved locations, cans (3.2%), Hispanic (1.3%), and other self-identified racial groups
previous destinations, and routes). Additionally, V2V data may represented (2.9%).
yield spatialetemporal positioning of the crash vehicle in relation In the Southern state, a total of 168 individuals participated from
to otherwise non-involved cars on the road, thus generating the two agencies responsible for responding to and investigating
“vehicle witnesses” (Fuentes et al., 2009; Kopylova et al., 2011). fatal traffic accidents, which reflected a response rate of 25.3% of all
Such electronic data can significantly supplement physical evi- active sworn forces (n ¼ 664). Within the sample, 97.02% of re-
dence collected from the crime scene, as skid marks and debris may spondents were male and 82.74% were white, which is in keeping
go unnoticed, be misattributed, or otherwise contaminated by in- with the general distribution in the overall force (96.3% and 81.1%,
vestigators and emergency responders on-scene (Boon, 2014). respectively). In terms of rank, 48.21% explicitly identified their
The sheer quantity of digital evidence made available within rank/pay level as trooper or senior trooper (sr./troopers represent
vehicles should be a boon to the officers who must investigate fatal 77.8% of the total population in this study), 13.69% identified their
auto accident scenes, particularly in determinations of driver rank/pay level as Special Agent or Senior Special Agent (22.4% of the
distraction and vehicle system failures. There is, however, little total population in this study), and 38.69% identified their rank as
research considering the extent to which first responders are aware other with either organization (e.g., Captain, Corporal, Investi-
of digital evidence as materials worthy of capture on the scene. gator). In addition, 78.99% were assigned to a highway patrol/traffic
Limited criminal justice research suggests police officers and first detail, meaning respondents were more likely to act as first re-
responders have mixed awareness of cybercrimes generally, and sponders to vehicular crash scenes. The average age of respondents
technologies more broadly (Goodison et al., 2015; Holt et al., 2010, was 35e54 years old (measured as a categorical variable), which
2015; Stambaugh et al., 2001). In addition, larger jurisdictions tend closely matched the overall population average of troopers and
to have the necessary resources and manpower needed to staff agents from the Southern state (42.5 years old).
specialized digital evidence-related units (Holt et al., 2010; Willits The final sample consisted of 450 respondents due to listwise
and Nowaki, 2016). deletion. There were 302 respondents from the Midwest (67.1%)
In the absence of specialized staff to seize digital evidence, the and 148 from the South (32.9%), which was proportionate to the
responsibility falls to the first responder at the crime scene to size of the respective state agencies in general. In fact, the force
identify and minimally triage the devices that may support of an strength of the Midwestern state police force was more than double
investigation (Goodison et al., 2015; Hinduja, 2007; NIJ, 2008). that of the Southern state. As a result, this convenient, yet purpo-
Recently, the International Association of Chiefs of Police (2019) sive sample from both states appeared to reflect the larger popu-
noted the lack of digital evidence training provided to police lation of the state police forces generally. Thus, this population
academy recruits and emphasized the need for improved delivery appeared appropriate for use in this exploratory analysis.
of content through continuing education for police staff. A range of
field guides have been produced (IACP 2019; NIJ, 2008), but little 3.1. Dependent variables
research considers how often the average first responder applies
this knowledge in the field (except Lee et al., 2019). Should a police To assess officers’ ability to recognize digital evidence in the
officer fail to properly secure devices at the scene or mishandle field, survey respondents were presented with a vignette
equipment, it may be possible to challenge the evidence later describing a call for service that they may initially encounter on
produced at trial and complicate a prosecution (Dolliver et al., 2017; duty. Scenarios are common in criminological and political science
Hinduja, 2007; Stambaugh et al., 2001). research to assess individual actions in unique situations (Fishbein
The unique challenges posed by the nature of digital evidence and Ajzen, 1975; Higgins et al., 2005; Lupariello et al., 2018; Piquero
may complicate the process of identification, search, and seizure by and Bouffard, 2007). Additionally, scenarios appear to be valid
first responders in the field. Thus, this comparative study examined mechanisms to assess actual behaviour in the real world (Fishbein
the factors associated with digital evidence recognition at fatal and Ajzen, 1975; Kim and Hunter, 1993; Lupariello et al., 2018).
vehicle crash scenes through the application of a unique survey To that end, a vignette was designed with input from both state
instrument completed by respondents in two state police agencies. police agencies to present a situation to respondents reflecting real
experiences in the field. The following language was used:
3. Data and methods You respond to the scene of a two-car traffic crash, which ap-
pears to have had one fatality, killing the driver of the second
Data for this study were collected through a survey instrument vehicle. Witnesses indicated that the driver of the first vehicle was
distributed to all sworn officers and troopers within a large Mid- driving erratically. Upon interacting with the driver of the first
western state police force and all sworn troopers and State Bureau vehicle, you note that they do not appear intoxicated. With the field
of Investigation (SBI) Special Agents in a large Southern state police interview of driver one, what items, if any, will you collect for po-
force. An email survey was distributed via internal mailing systems tential evidentiary value? Please select all that you think apply.
to both police forces in February 2019. The initial email solicitation After this description, respondents were presented with the
included a description of the study, the ethical protections provided following question: “With the field interview of driver one, what
to participants, and a link for the electronic survey instrument. items, if any, will you collect for potential evidentiary value? Please
Participation was voluntary and no enticements were provided to select all that you think apply.” Participants were presented with
respondents. A follow-up request was distributed in March, 2019, four total response options: 1) mobile devices; 2) fast food that may
and data collection terminated in April, 2019. have been within reach of the driver; 3) makeup and/or hair
A total of 345 individuals participated in the Midwestern state, products that may have been within reach of the driver; 4) vehicle
though there was some variation in complete responses. For infotainment or telematics information; and 5) other. The scenario
3
T. Holt and D.S. Dolliver Forensic Science International: Digital Investigation 37 (2021) 301167

was designed to elicit multiple responses from participants on the of respondents (87.2%) identifying as white (see Table 1). State of
basis of legal precedents, recommendations from professional or- origin was included as a binary to reflect the Midwest (0) or South
ganizations, and field experience. In addition, the statement what (1). Lastly, gender was measured as a binary variable (0 ¼ male;
has “potential evidentiary value” was deliberate so as to allow re- 1 ¼ female).
spondents to identify any and all materials they felt would be
relevant. Thus, the selection of digital and physical materials was 4. Results
allowed, so long as they felt it would support their investigation in
the scenario. Respondents were asked to assess their comfort level with
The responses to the scenario were used as dependent variables computer use to assess what impact technology use may have on
in this analysis to identify any significant relationships between digital evidence recognition (e.g. Holt and Bossler, 2012; Holt et al.,
respondents’ views and various occupational and demographic 2019). The majority of respondents reported being either
measures as independent variables. Multiple tests of significance comfortable (n ¼ 146; 32.6%) or very comfortable (n ¼ 160; 35.7%)
were utilized (chi-square, t-tests, Spearman correlations), and any using computers every day. A small proportion also noted they
significant relationships observed were highlighted using asterisks were very uncomfortable using computers (n ¼ 52; 11.6%) in
to denote the level of significance. keeping with prior research among line officers in police agencies
(Holt and Bossler, 2012; Holt et al., 2019). Those who received
3.2. Independent variables training in digital forensics evidence recovery and response were
significantly more likely to be comfortable using a computer
Four measures were included to assess any potential relation- (c2 ¼ 24.635***; t ¼ 4.779***).
ship between respondents’ occupational experiences and their A substantial proportion of respondents (n ¼ 185; 41.3%) also
knowledge of digital evidence (see Table 1 for detail). These items received training in investigating online incidents and crimes. Those
were included based on relationships identified in prior studies on with fewer years of experience were less likely to have received
police perceptions of cybercrimes and crimes with digital eviden- training (c2 ¼ 13.416*,.t ¼ 2.277*), while those who were younger
tiary components (e.g. Hinduja, 2004; Holt and Bossler, 2012; (t ¼ 1.956*), from the Midwest (c2 ¼ 20.358***,.t ¼ 4.756***), and
Senjo, 2004). First, participants were asked to assess their comfort had more education were significantly more likely to have received
level with computer use through a six-item Likert scale (1 ¼ very training (c2 ¼ 11.382*,.t ¼ 3.225***). Additionally, most respondents
uncomfortable; 2 ¼ uncomfortable; 3 ¼ somewhat uncomfortable; knew who to contact in the event they needed assistance regarding
4 ¼ somewhat comfortable; 5 ¼ comfortable; 6 ¼ very comfort- digital evidence recovery while at the scene of a crime (n ¼ 357;
able). Second, respondents were asked when was the last time they 86.9%). Those who were younger (c2 ¼ 11.269*, t ¼ 2.778**), from
responded to a crime scene where digital evidence was present the Midwest (c2 ¼ 15.268***, t ¼ 3.601***), and were more
using a six item response (digital evidence scene: 1 ¼ never; 2 ¼ over comfortable using a computer (x2 ¼ 11.970*; t ¼ 1.932*) were more
a year ago; 3 ¼ within the last year; 4 ¼ within the last several likely to know their digital evidence contact.
months; 5 ¼ within the last few weeks; 6 ¼ within the last week). Respondents were also asked to assess the last time they
Two additional binary variables were included: whether they responded to a scene where digital evidence was present. A sub-
received training in the investigation of online crime, and if the stantive proportion responded to such a scene within the last week
officer knew their digital evidence contact (0 ¼ no; 1 ¼ yes). (n ¼ 104; 23.9%), or the last few weeks (n ¼ 79; 18.2%). A small
Additionally, seven variables were included as demographic proportion responded to such a scene over a year ago (n ¼ 74; 17%),
controls, including age (1 ¼ 18e24; 2 ¼ 25e34; 3 ¼ 35e44; or never responded to a scene with digital evidence in their career
4 ¼ 45e54; 5 ¼ 55e64; 6 ¼ 65 and older), education (1 ¼ high (n ¼ 49; 11.3%). White respondents were more likely to have
school; 2 ¼ some college; 3 ¼ college degree; 4 ¼ some graduate responded to a scene in the last year (c2 ¼ 19.356**; t ¼ 2.885**),
school; 5 ¼ graduate degree), and years of policing experience as were those who received training (c2 ¼ 55.270***; t ¼ 8.220***),
(1 ¼ less than one year; 2 ¼ 1e5 years; 3 ¼ 6e10 years; 4 ¼ 11e14 were from the Midwest (c2 ¼ 50.436***, t ¼ 6.495***), knew
years; 5 ¼ 15e20 years; 6 ¼ 21e30 years; 7 ¼ 31 or more years). their digital evidence contact (c2 ¼ 44.154***; t ¼ 4.220***), and
These items were measured categorically to simplify the response were more comfortable using a computer (t ¼ 2.695**).
options for participants. The majority of respondents were between There were two primary responses among respondents to the
45 and 54 years of age (n ¼ 149; 35.9%) and reported having 21e30 scenario question regarding what should be done at the scene of a
years of experience (n ¼ 126; 30.1%). Rank was included as a binary fatal accident: 1) collect mobile devices (86.9%) and 2) the tele-
measure that segmented respondents into troopers, officers, or matics of the vehicle (73.6%; see Table 2 for detail). This distribution
investigators (0) and sergeants (1). Race was included as a binary is sensible given the potential information that may be stored on
measure (0 ¼ white; 1 ¼ minority group member) due to a majority these devices to assist in the investigation and reconstruction of the
accident. A proportion of respondents also indicated they would
search for makeup (24.9%) and fast food (24.2%) around the drivers,
Table 1 which indicates a lack of understanding in terms of what
Descriptive statistics of sample population.

Variable Mean Min Max


Table 2
Comfort Level (n ¼ 448) 4.64 1 6 Respondents’ decisions regarding what actions to take in the digital evidence sce-
Digital Evidence Scene (n ¼ 435) 3.85 1 6 nario (n ¼ 450).
Training (0 ¼ no, n ¼ 448) 50.8% Yes 0 1
Digital Evidence Contact (n ¼ 434) 82.2% Yes 0 1 Response Options Full Sample Midwestern Southern
Age (n ¼ 415) 2.70 1 6
Percent N Percent N Percent N
Education (n ¼ 421) 2.70 1 5
Years of Police Experience (n ¼ 418) 4.39 1 7 Mobile device 86.9 391 92.1 278 76.4 113
Rank (n ¼ 414) 70.6% Trooper 0 1 Fast food 24.2 109 20.2 61 32.4 48
Race (n ¼ 414) 87.2% White 0 1 Makeup 24.9 112 20.9 63 33.1 49
State (n ¼ 450) 67.2% Midwest 0 1 Telematic 73.6 331 74.2 224 72.3 107
Gender (n ¼ 416) 86.5% Male 0 1 Other 16.0 72 17.5 53 12.8 19

4
T. Holt and D.S. Dolliver Forensic Science International: Digital Investigation 37 (2021) 301167

constitutes objects with evidentiary value. evidence were more likely to properly identify and seize such
A small number of respondents also noted the other category materials at the scene of fatal vehicle crashes. Across both states,
(16%), with differing perspectives on what would need to be done at officers and agents overwhelmingly recognized the need to recover
the scene. Twenty respondents (4.4%) specifically stated the need to mobile devices and telematics from the vehicle in the vignette.
collect the car's crash data retrieval (CDR) or electronic control unit Additionally, a substantive number of officers remained unsure of
(ECU), as this will contain information about vehicle operation at the potential digital evidentiary value of other objects on-scene
the time of the crash. Sixteen respondents also noted the need to (Rivers, 2011; Van Kirk, 2000). Approximately one-quarter of re-
photograph the contents of the vehicle at the time of the crash. A spondents indicated that they would search for makeup and fast
small proportion (n ¼ 12; 2.6%) also noted they would search the food from the accident scene. Those respondents were more likely
vehicle for GPS systems, or simply impound the entire vehicle to be older or have more years of experience respectively, sug-
(n ¼ 7; 1.5%). gesting that there may be generational differences at play when
Some unique relationships were identified with respect to the considering what evidence may have value to an investigation (Holt
items that should be searched and the demographics or experi- et al., 2019).
ences of respondents. For instance, females were more likely to Officers with fewer years of experience on the job were also less
identify mobile devices (c2 ¼ 4.309*; t ¼ 4.848***), as were those likely to have received training in investigating online incidents and
who received training (c2 ¼ 6.045*; t ¼ 2.599**), knew their digital crimes with digital components. This may have occurred organi-
evidence contact (c2 ¼ 20.616***; t ¼ 3.286***), and had more cally within the agencies, as less time on the force objectively yields
recently responded to a scene with digital evidence fewer opportunities for training. This relationship may also indicate
(c2 ¼ 21.179***; t ¼ 3.781). a preference in agencies selectively opting to send more senior
Individuals who knew their digital evidence contact were more members for training over newer hires (see Holt et al., 2015;
likely to say no to collecting fast food (c2 ¼ 6.599**; t ¼ 2.319*) Stambaugh et al., 2001). The latter is potentially problematic, as
and makeup (c2 ¼ 5.723*; t ¼ 2.180*). By contrast, those with more senior members of the agencies are often at higher ranks
more years of experience were more likely to collect fast food within the organization and are closer to retirement. Such in-
(t ¼ 2.428*), and makeup (t ¼ 2.985). Older respondents dividuals may also be less likely to serve as a first responder at a
(c2 ¼ 14.923**; t ¼ 3.079**) were also more likely to collect fatal vehicle crash, which could minimize the likelihood of proper
makeup. Those who were from the large Southern state were also identification and collection of digital evidence on-scene (Goodison
more likely to collect fast food (c2 ¼ 8.099**; t ¼ 2.714**) and et al., 2015; Holt et al., 2015).
makeup (c2 ¼ 7.969**; t ¼ 2.698**). As a result, this research underscores the necessity for agencies
Those who responded to a scene with digital evidence more in the United States to improve digital evidence training across all
recently were more likely to note the need to collect other materials ranks and time on the force for those individuals responding to fatal
(c2 ¼ 14.677*; t ¼ 3.828***), along with those who received digital traffic accidents and other calls for service (see also Holt et al., 2015;
forensics training (c2 ¼ 7.197**; t ¼ 2.597**), white officers Kennedy et al., 2019). Indeed, the International Association of
(c2 ¼ 3.640*; t ¼ 2.427*), and those with higher education levels Chiefs of Police (2019) recognized the lack of digital evidence
(t ¼ 2.635**). Taken as a whole, these findings suggest that personal training among police academy recruits and called for in-service
experiences and understanding of digital forensics may be a key trainings related to digital evidence identification and recovery. It
factor to increase responders’ recognition of digital evidence at fatal is unlikely that such a program could be implemented nationwide
crashes in the United States (see also Holt et al., 2019). While these due to the inconsistencies observed across state budgets for
findings are particular to agencies in two U.S. states, the findings training, particularly around specialized issues like digital evidence
lend insight to potential broader trends nationally and handling and collection (Dolliver, 2019; Dolliver et al., 2017; Holt
internationally. et al., 2015).
There may be greater value in implementing consistent digital
5. Discussion and conclusions and thorough evidence training during academy training for cadets
and recruits in state police agencies (IACP 2019; Stambaugh et al.,
Digital forensic evidence has become an essential component of 2001). Early exposure to accurate and efficient collection methods
criminal investigations across the United States at all levels of law for digital evidence while on-scene could establish baseline
enforcement (Casey, 2011; Dolliver, 2019; Holt and Bossler, 2015; knowledge within generations of state police and improve the
Goodison et al., 2015; Lupariello et al., 2018; Stambaugh et al., overall response to fatal accidents as the number of smart vehicles
2001). The use of these materials will only become more com- increases over time (see also Kennedy et al., 2019). Since digital
mon as technologies continue to evolve, and society becomes even evidence skills also transfer to all criminal case-types, early career
more dependent on Internet-enabled devices (Dolliver, 2019; training could improve the long-term capacity of state police
Tuner, 2019). Law enforcement officers at the line level in local and agencies to produce quality evidence for all manner of in-
state agencies are expected to accurately identify and secure digital vestigations, whether in physical or digital spaces (Dolliver, 2019;
devices for later forensic capture by specialized units, individuals, Hinduja, 2007; Holt et al., 2015; Stambaugh et al., 2001).
or task forces (Dolliver et al., 2017; Hinduja, 2007; Holt et al., 2015; Differences were also observed across respondents on the basis
Stambaugh et al., 2001). However, limited evidence suggests line of their state with regard to experience and preparedness to handle
officers are not comfortable with the practice or processes involved digital forensic evidence. While the demographic composition of
in digital forensic evidence identification and collection (Holt and the two state agencies included in this study were relatively com-
Bossler, 2012; Holt et al., 2010, 2015; Willits and Nowacki, 2016). parable aside from their overall size, there were significant opera-
As such, this study sought to address this gap in literature by tional differences in digital forensic capabilities. Officers from the
surveying state law enforcement officers and agents responsible for Midwest were more likely to know who to contact in the event they
investigating such calls for service from a large Midwestern state needed assistance regarding digital evidence recovery while at the
and a large Southern state in the U.S. scene of a fatal crash, receive digital evidence and cyber-related
There are several key implications from this analysis, notably case training, and were more comfortable using a computer. Mid-
that training affects one's recognition of digital evidence. Re- western officers were also more likely to be college-educated than
spondents with prior training and field experiences with digital their Southern counterparts. The majority of Southern officers
5
T. Holt and D.S. Dolliver Forensic Science International: Digital Investigation 37 (2021) 301167

recognized the need to seize mobile devices in the scenario, but References
they were also more likely to collect fast food and makeup.
These findings call to question what factors shape differences in Anderson, M., 2019. Mobile Technology and Home Broadband, 2019. Pew Internet
and American Life Project. https://www.pewinternet.org/2019/06/13/mobile-
the capacities of state agencies. It is possible that the data reflect technology-and-home-broadband-2019/.
differential impacts of budget size and resource allocations for in- Anderson, M., Jiang, J., 2018. Teens' Social Media Habits and Experiences. Wash-
service training and recruitment, as well as state-level variation ington, DC: Pew Internet and American Life Project. Retrieved December, 6,
2018.
in training opportunities for digital forensics in general (Holt et al., Boon, R., 2014. Post-accident analysis of digital sources for traffic accidents. 21st
2015; Stambaugh et al., 2001; Willits and Nowacki, 2016). Addi- Twente Student Conference on IT, Enschede, The Netherlands. Accessed via.
tional research is needed to replicate this study using multiple state https://pdfs.semanticscholar.org/cb7b/
7a83d18ceb66c7c1cfc74f5160a840c652f5.pdf?_ga¼2.25389625.564902324.
police agencies across the U.S. to better assess the factors affecting 1578168814-605327610.1578168814.
officer competencies (Holt et al., 2019; Lupariello et al., 2018). Casey, E., 2011. Digital Evidence and Computer Crime: Forensic Science, Computers
The comparative nature of this study also limits it generaliz- and the Internet, third ed. Elsevier, Waltham, MA.
Chan, M., Singhal, A., 2015. Emotion matters: implications for distracted driving. Saf.
ability to other populations. The response rates for the populations
Sci. 72, 302e309.
from the two states (18% and 25%, respectively) were lower than DeAngelo, Gregory, Hansen, Benjamin, 2014. Life and Death in the fast lane: police
expected despite multiple follow-up survey reminders. Lower Enforcement and traffic fatalities. Am. Econ. J. Econ. Pol. 6 (2), 231e257.
Dimitrakopoulos, G., 2011 August. Intelligent transportation systems based on
response rates in survey-based policing research are not uncom-
internet- connected vehicles: fundamental research areas and challenges. In: In
mon (Groves and Peytcheva, 2008; Nix et al., 2017), and it is ITS Telecommunications (ITST), 2011 11th International Conference on, vols.
possible that respondents may have purposively avoided taking the 145e151. IEEE.
survey or answered questions differently due to various individual Dolliver, Diana S., 2019. Emerging technologies, law enforcement responses, and
national security. I/S. A Journal of Law and Policy for the Information Society 15
factors. The use of a scenario-based measure also merits further (1e2), 123e150.
examination, as the conditions may not fully reflect those experi- Dolliver, D.S., Collins, C., Sams, B., 2017. Hybrid approaches to digital forensic in-
enced in the field. The use of qualitative field studies with local and vestigations: a comparative analysis in an institutional context. Digit. Invest.:
The International Journal of Digital Forensics & Incident Response 23 (4),
state police at traffic accident calls for service could be essential to 124e137.
better document the ways that evidence is triaged in real time Ferguson, G., 2017. The Rise of Big Data Policing: Surveillance, Race, and the Future
(Hinduja, 2007; Holt et al., 2015). Such research could also benefit of Law Enforcement. NYU Press, NY.
Fishbein, M., Ajzen, I., 1975. Intention and Behavior: An Introduction to Theory and
our understanding of the extent to which evidence may be pulled Research.
from the vehicles involved in accidents relative to the personal Fuentes, J.M., Gonzalez-Tablas, A.I., Ribagorda, A., 2009. Witness-based evidence
devices of the drivers and passengers involved. Differences in the generation in vehicular ad-hoc networks. In: Embedded Security in Cars Con-
ference (ESCAR). Accessed via. https://pdfs.semanticscholar.org/f7da/
digital capacities of vehicles on the road at any time may also limit
feb9fc89a23bd234cf44cd104325fdee6d65.pdf?_ga¼2.260378121.564902324.
their ability to retain evidence, creating variations in the ways law 1578168814-605327610.1578168814.
enforcement can utilize vehicular data collected from scenes Gerla, M., Lee, E.-K., Pau, G., Lee, U., 2014 March. Internet of vehicles: from intelli-
gent grid to autonomous cars and vehicular clouds. In: In Internet of Things
(Kennedy et al., 2019). Thus, future research is essential to disen-
(WF-IoT), 2014 IEEE World Forum on. IEEE, Seoul, South Korea, pp. 241e246.
tangle the variety of technological and human factors that may be Goodison, S.E., Davis, R.C., Jackson, B.A., 2015. Digital Evidence and the US Criminal
involved in fatal accident investigations over time. Justice System. Identifying Technology and Other Needs to More Effectively
Future research should also more thoroughly examine advanced Acquire and Utilize Digital Evidence. Priority Criminal Justice Needs Initiative.
Rand Corporation, San Diego, CA.
digital forensic capabilities and data acquisition techniques that Groves, R.M., Peytcheva, E., 2008. The impact of nonresponse rates on nonresponse
exist at the local and state level law enforcement agencies in the bias. Publ. Opin. Q. 72 (2), 167e189.
U.S. (Holt et al., 2015; Willits and Nowacki, 2016) Variations Higgins, G., Wilson, A., Fell, B., 2005. An application of deterrence theory to soft-
ware piracy. J. Crim. Justice Popular Cult. 12 (3), 166e184.
observed within and across states may impact the rate of successful Hinduja, S., 2004. Perceptions of local and state law enforcement concerning the
case processing, and limit capacity building across the nation. role of computer crime investigative teams. Polic. Int. J. Police Strat. Manag. 27
While police are the front line of criminal investigations, further (3), 341e357. https://doi.org/10.1108/13639510410553103.
Hinduja, S., 2007. Computer crime investigations in the United States: leveraging
research is also required to better understand how well prosecu- knowledge from the past to address the future. International Journal of Cyber
tors, judges, and defense attorneys understand digital forensic Criminology 1 (1), 1e26.
processes (Dolliver, 2019; Holt et al., 2015) Challenges in commu- Holt, T.J., Bossler, A.M., 2012. Police perceptions of computer crimes in two south-
eastern cities: an examination from the viewpoint of patrol officers. Am. J. Crim.
nications and comprehension of issues related to evidence collec-
Justice 37 (3), 396e412.
tion, preservation, and analysis may significantly affect trial Holt, T.J., Bossler, A.M., 2015. Cybercrime In Progress: Theory and Prevention Of
outcomes. Finally, this study should be replicated in an interna- Technology-Enabled Offenses. Routledge, New York.
Holt, T.J., Bossler, A.M., Fitzgerald, S., 2010. Examining state and local law enforce-
tional context to more fully understand the scope and depth of
ment perceptions of computer crime. Crime on-line: Correlates, causes, and
digital forensic knowledge; as societies continue to become ever- context 221e246.
more digitally connected, the volume and types of digital evi- Holt, T.J., Burruss, G.W., Bossler, A., 2015. Policing Cybercrime and Cyberterror.
dence will only continue to grow and impact law enforcement Carolina Academic Press, Raleigh, NC.
Holt, T.J., Clevenger, S., Navarro, J., 2019. Exploring digital evidence recognition
agencies in countries around the world. among officers and troopers in a sample of a state police force. Policing: Int. J.
43 (1).
Horsman, G., Conniss, L., 2015. Investigating evidence of mobile phone usage by
drivers in road traffic accidents. Digit. Invest. 12, 30e37.
Conflict of interest 2019. IACP. (Accessed 3 May 2021).
International Chiefs of Police, 2019. Understanding digital evidence. Accessed at.
http://www.iacpcybercenter.org/investigators/digital-evidence/understanding-
The authors have no conflict of interest to report. digital-evidence/.
Kennedy, J., Holt, T., Cheng, B., 2019. Automotive cybersecurity: assessing a new
platform for cybercrime and malicious hacking. J. Crime Justice 42 (5),
632e645.
Kim, M.S., Hunter, J.E., 1993. Relationships among attitudes, behavioral intentions,
Acknowledgements
and behavior: a meta-analysis of past research, part 2. Commun. Res. 20 (3),
331e364.
The authors would like to thank the law enforcement agencies Kopylova, Y., Farkas, C., Wenyuan, X., 2011. Accurate accident reconstruction in
VANET. Data and Applications Security and Privacy 6818, 271e279.
in the two states included in this study for their cooperation and
Lee, J.R., Holt, T.J., Burruss, G.W., Bossler, A.M., 2019. Examining English and Welsh
participation in this research project.
6
T. Holt and D.S. Dolliver Forensic Science International: Digital Investigation 37 (2021) 301167

detectives views of online crime. Int. Crim. Justice Rev. https://doi.org/10.1177/ prevalence, problems, and prevention. Int. J. Inj. Contr. Saf. Promot. 22 (3),
1057567719846224. 187e192.
Lupariello, F., Curti, S., Duval, J.B., Abbattista, G., Vella, G.D., 2018. Staged crime scene Piquero, A.R., Bouffard, J.A., 2007. Something old, something new: a preliminary
determination by handling physical and digital evidence: reports and review of investigation of Hirschi's redefined self-control. Justice Q. JQ 24 (1), 1e27.
the literature. Forensic Sci. Int. 288, 236e241. https://doi.org/10.1016/ Rivers, R.W., 2011. Traffic Accident Investigators' Handbook, third ed. Charles C.
j.forsciint.2018.04.050. Thomas Publishers, Springfield, IL.
Maras, M.H., 2012. Computer Forensics: Cybercriminals, Law, and Evidence. John & Schmelzer, R., 2019. What happens when self-driving cars kill people? Forbes.
Bartlett Learning, LLC, New York, NY. September 26, 2019. https://www.forbes.com/sites/cognitiveworld/2019/09/26/
National Center for Statistics and Analysis, 2019. January. Driver Electronic Device what-happens-with-self-driving-cars-kill-people/#472738c3405c.
Use in 2017 (Traffic Safety Facts Research Note. Report No. DOT HS 812 665). Senjo, S., 2004. An analysis of computer-related crime: comparing police officer
National Highway Traffic Safety Administration, Washington, DC. perceptions with empirical data. Sec. J. 17, 55e71.
National Highway Traffic Safety Administration (NHTSA), 2019. Traffic Safety Facts. Stambaugh, H., Beaupre, D.S., Icove, D.J., Baker, R., Cassaday, W., Williams, W.P.,
U.S. Department of Transportation DOT HS 812 783. National Center for Sta- 2001. Electronic Crime Needs Assessment for State and Local Law Enforcement
tistics and Analysis, Washington, D.C. Series. NCJ. Retrieved from. https://www.ncjrs.gov/txtfiles1/nij/186276.txt.
National Institute of Justice (NIJ), 2008. Electronic Crime Science Investigation: A Tene, O., 2008. What google knows: privacy and internet search engines. Utah Law
Guide for First Responders, second ed. U.S. Department of Justice, National Rev. 1433.
Institute of Justice, Washington, DC. Tuner, J.L., 2019. Managing digital discovery in criminal cases. J. Crim. Law Criminol.
Nilsson, D., Phung, P., Larson, U., 2018. Vehicle ECU classification based on safety- 109 (2), 237e311.
security characteristics. IEEE Xplore. https://doi.org/10.1049/ic.2008.0810. Van Kirk, D.J., 2000. Vehicular Accident Investigation and Reconstruction. Crc Press.
Nix, J., Pickett, J., Baek, H., Alpert, G., 2017. Police research, officer surveys, and Walker, S., Katz, C.M., 2012. Police in America. McGraw-Hill.
response rates. Polic. Soc. 29 (5), 530e550. Willits, D., Nowacki, J., 2016. The use of specialized cybercrime policing units: an
Novak, M. A. R. T. I. N., Grier, J., Gonzales, D. A. N. I. E. L., 2018. New approaches to organizational analysis. Crim. Justice Stud. Crit. J. Crime Law Soc. 29 (2),
digital evidence acquisition and analysis. NIJ Journal No 280. 105e124.
Oulasvirta, A., Rattenbury, T., Ma, L., Raita, E., 2012. Habits make smartphone use Wilson, F.A., Stimpson, J.P., 2010. Trends in fatalities from distracted driving in the
more pervasive. Personal Ubiquitous Comput. 16 (1), 105e114. United States, 1999 to 2008. Am. J. Publ. Health 100 (11), 2213e2219.
Overton, T.L., Rives, T.E., Hecht, C., Shafi, S., Gandhi, R.R., 2015. Distracted driving:

You might also like