National Academies Press: OpenBook

E-Scooter Safety Toolbox (2023)

Chapter: Chapter 4 - Data Tools and Methods

« Previous: Chapter 3 - Promising Practices
Page 15
Suggested Citation:"Chapter 4 - Data Tools and Methods." National Academies of Sciences, Engineering, and Medicine. 2023. E-Scooter Safety Toolbox. Washington, DC: The National Academies Press. doi: 10.17226/27253.
×
Page 15
Page 16
Suggested Citation:"Chapter 4 - Data Tools and Methods." National Academies of Sciences, Engineering, and Medicine. 2023. E-Scooter Safety Toolbox. Washington, DC: The National Academies Press. doi: 10.17226/27253.
×
Page 16
Page 17
Suggested Citation:"Chapter 4 - Data Tools and Methods." National Academies of Sciences, Engineering, and Medicine. 2023. E-Scooter Safety Toolbox. Washington, DC: The National Academies Press. doi: 10.17226/27253.
×
Page 17
Page 18
Suggested Citation:"Chapter 4 - Data Tools and Methods." National Academies of Sciences, Engineering, and Medicine. 2023. E-Scooter Safety Toolbox. Washington, DC: The National Academies Press. doi: 10.17226/27253.
×
Page 18
Page 19
Suggested Citation:"Chapter 4 - Data Tools and Methods." National Academies of Sciences, Engineering, and Medicine. 2023. E-Scooter Safety Toolbox. Washington, DC: The National Academies Press. doi: 10.17226/27253.
×
Page 19
Page 20
Suggested Citation:"Chapter 4 - Data Tools and Methods." National Academies of Sciences, Engineering, and Medicine. 2023. E-Scooter Safety Toolbox. Washington, DC: The National Academies Press. doi: 10.17226/27253.
×
Page 20
Page 21
Suggested Citation:"Chapter 4 - Data Tools and Methods." National Academies of Sciences, Engineering, and Medicine. 2023. E-Scooter Safety Toolbox. Washington, DC: The National Academies Press. doi: 10.17226/27253.
×
Page 21
Page 22
Suggested Citation:"Chapter 4 - Data Tools and Methods." National Academies of Sciences, Engineering, and Medicine. 2023. E-Scooter Safety Toolbox. Washington, DC: The National Academies Press. doi: 10.17226/27253.
×
Page 22
Page 23
Suggested Citation:"Chapter 4 - Data Tools and Methods." National Academies of Sciences, Engineering, and Medicine. 2023. E-Scooter Safety Toolbox. Washington, DC: The National Academies Press. doi: 10.17226/27253.
×
Page 23

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

15   Data Tools and Methods Using Existing Injury Data Sources To evaluate programs and adjust policy approaches, agencies need crash and injury data and evidence to inform conversations and engage community members who may be able to contex- tualize such information. Ideally, these data would be readily available, consistently gathered and shared in a timely fashion, and comprehensive and accurate. Data needed for e-scooter injury measurement are more often inaccessible, miscoded or missing, and inconsistently reported across place or time. Agencies do not need perfect data systems to help inform good decisions, but understanding existing data and their limitations can be helpful. Table 4 shows four different examples of com- monly available sources of crash and injury data that agencies can use to inform their program development and monitor outcomes. There are many known limitations and biases in the data systems shown in Table 4. For example, children and older adults may be more likely to seek and receive hospital care, while men and people (people other than white men) aged 20–30 may be underrepresented in health care datasets (Harmon et al. 2021). Injuries affecting migrant populations, communities of color, and indigenous groups may be under- or overrepresented in health care and police crash datasets because of racism and other social, environmental, and economic conditions that impact care-seeking and incident reporting (Doggett et al. 2018; National Academies of Sciences, Engineering, and Medicine 2018). It is essential that agencies seeking to monitor and evaluate their safety programs use sources beyond standard police and health care records. Community data and input on safety (e.g., qual- itative data) are critical. There are many ways community members can provide meaningful feedback on an ongoing basis to e-scooter program safety managers. Common methods include: • Surveys, such as sidewalk intercept surveys, phone surveys, or web-based surveys • In-app feedback • Focus groups • Listening sessions • Community e-scooter rides and audits • Town halls and other meetings or pop-up venues for community input C H A P T E R 4 Table 2 provides additional resources for community-engaged data collection and input that can supplement existing injury and crash data sources. Table 4 provides a list of core data needs for agencies to use to make e-scooter safety measurement more consistent, reliable, timely, and accessible.

16 E-Scooter Safety Toolbox Data Source How can I access it? What can it tell me and what should I be aware of? Police crash records • State department of motor vehicles • PedBikeData (CSCRS 2022) • Most states lack an e-scooter code; incidents could be coded as motorcyclist or pedestrian or something else; must search narrative text field to identify cases. • Usually excludes crashes that don’t involve a motor vehicle or a threshold of damage (which may represent a large proportion of e-scooter crashes). • May exclude crashes that occur off-road, including sidewalks, parking lots, and trails. • Published/accessible data can lag 1-2 years behind. Hospital/ emergency department visit records • State/local health departments • State syndromic surveillance systems and/or discharge data systems • Captures injuries to bystanders and riders, as well as injuries due to falls and motor vehicle collisions. • There are standardized case definitions of e-scooter injuries (ICD-10-CM codes available as of October 2020; see glossary). • Data about patient, injury mechanisms, and injury diagnoses. • Depending on the data source, data may be available in near real time. • Contains indirect measures of injury severity, such as hospital/emergency department discharge disposition and hospital length of stay. • Limited information about crash location/context. Trauma registry records • Directly through local trauma center • Trauma registries (often run by state departments of health or emergency medical services (EMS)) • Contains ICD-10-CM external cause of injury codes for the classification of e-scooter injuries. • Captures injuries to bystanders and riders, as well as injuries related to falls and motor vehicle collisions. • Contains direct measures of injury severity, such as the Abbreviated Injury Scale, the Injury Severity Score, and other injury severity metrics. • Provides quality data about patient and injury outcomes, but limited information about crash location/context. • Data are limited to the most severe cases, in which the patient dies, is transferred from another facility, or requires hospitalization. • Some states have few trauma centers. EMS run reports • National Emergency Medical Services Information System (for de-identified national data) • State or local offices of emergency management/services (for geocoded data) • Office of Emergency Management Systems (for de-identified data) • Contains ICD-10-CM external cause of injury codes for the classification of e-scooter injuries. • Captures injuries to bystanders and riders, as well as injuries related to falls and motor vehicle collisions. • Quality data on EMS activation/travel times and procedures administered. • Data on incident location (including geolocation for some agencies), primary patient impression, patient symptoms, mechanism of injury, and location of injury. • Local data may include a narrative patient care report, providing additional context. • Local data may be available in near real time. • Only a fraction of incidents involves EMS. • Limited information on patient outcomes and injury diagnoses. • Highly variable in data quality and standardized elements. (NEMSIS) Table 4. Sources of e-scooter crash and injury data. Gathering Additional Data for E-Scooter Safety Assessments Common mantras heard in transportation are that “what matters gets measured,” and “if you’re not counted, you don’t count.” Underreporting of injuries, crashes, and other risks experienced by e-scooter riders can contribute to underinvestment in safety programs and infrastructure improvements. There is a need for those responsible for managing and maintaining state injury records to be aware of issues specific to e-scooter injury and crash reporting and to foster improve- ments in data collection and management systems.

Data Tools and Methods 17   The previous section (see Table 4) laid out common sources of e-scooter injury and crash data as well as strengths and limitations of each. Crash and injury data often are just the tip of the iceberg, and other information can be helpful in contextualizing these events and being able to measure rates of events over time or relative frequency of incidents or risks among different groups. The first column in Table 5 presents additional data elements that may clarify e-scooter safety issues. These include information about the trip itself, the environment in which the trip was taken, and the device used, as well as information about the e-scooter user and their percep- tions, behaviors, and interactions with others on the roads. Unfortunately, there is no singular source of all this information. Rather, it is typically gathered by local agencies in an ad hoc Table 5. Data needs and which collection methods can provide such data to augment crash and injury records. Data Elements (1) Intercept Survey (2) Web- Based Survey (3) Direct Manual Observation (4) Indirect Manual Observation (video recording + processing) (5) Indirect Automated Observation (automated counters, sensors, etc.) (6) Mobility- (firm provided data) Trip purpose Yes Yes No No No No Trip length / distance Yes Yes No No No Yes Trip duration / time spent riding Yes Yes No No No Yes Trip location/ route No No No No No Yes Roadway, lighting, traffic, and weather conditions No No Yes Yes No No, unless data are linked E-scooter device characteristics Yes Yes Possibly Possibly Possibly Yes E-scooter speed Self- reported Self- reported Directly measured Directly measured Directly measured, depending on tech Yes Rider demographics Yes Yes Possibly Possibly No Yes Rider characteristics (riding in group, carrying objects, etc.) Yes Yes Yes Yes No No Helmet use Self- reported Self- reported Directly measured Directly measured Directly measured, depending on tech Possibly, if firm gathers Rider interactions and conflicts with other road users Self- reported Self- reported Possibly Possibly No No Perceptions of safety Yes Yes Indirectly based on behaviors Indirectly based on behaviors No Possibly, if firm gathers Rider behaviors (signaling, gesturing, yielding, piggybacking, using devices, looking, dismounting, parking, etc.) Self- reported Self- reported Directly measured Directly measured No No

18 E-Scooter Safety Toolbox fashion. Columns numbered one through six in Table 5 show different methods available for collecting the data elements needed for a holistic evaluation of e-scooter safety. Table 2 offers a list of data elements that could be collected via intercept, web, or routine travel surveys as well as through community-led processes. Chapter 7 in BTSCRP Web-Only Document 5 offers an extensive review of the strengths and limitations of two methods (direct manual observation and indirect manual observation) pilot tested in this study. Leveraging Data Improvement Opportunities Quality data, needed for e-scooter safety assessment, likely have the following characteristics: • Consistent and reliable • Timely • Accessible • Integrated • Performance/outcome oriented These quality data characteristics are not just principles; in many cases they are recommended or required by federal agencies. For example, NHTSA’s Model Minimum Uniform Crash Criteria (MMUCC) and Crash Data Improvement Program (CDIP) aim to make data “timely, accurate, complete, uniform, integrated and accessible” (NHTSA 2022a) in order to benefit a broad spec- trum of data users. The following sections describe common data limitations and opportunities to improve them to advance e-scooter safety data quality, with examples from cities documented in case studies and elsewhere. Consistent and Reliable Currently, police-reported e-scooter crash and operator-reported incident data are not reliable as they are not collected in consistent ways and most agencies do not provide rigorous and routine training for staff to code incidents. While police-reported crash data contain a variable from the MMUCC abbreviations (known as KABCO) (FHWA 2022c) indicating injury severity level, injury surveillance is not the primary function of crash data, and police officers generally are not trained diagnosticians. In contrast, there are standardized case definitions used in health care settings to code e-scooter related injuries, using the International Statistical Classification of Diseases and Related Health Problems (ICD), 10th Revision, Clinical Modification (ICD-10-CM). Additionally, clinicians, medical coders, trauma registrars, and others involved with health care data receive regular training to apply ICD-10-CM codes in a consistent and reliable way as part of their professions. Although it is facility dependent, most hospitals are likely collecting these data as part of their routine, administrative capacity and have established incentives to improve coding quality and consistency. Unlike health care data, which relies on nationally adopted standards (ICD-10-CM) for disease and injury classification (including e-scooter injuries), police-reported crash data are managed at the state level. Traffic Records Coordinating Committees (TRCC) aim to improve traffic safety data collection, management, and analysis at state and federal levels. Their work in safety data improvement covers crash and injury data and data related to the involved parties, vehicles, roadway context, and citation or court related data. Each state has a TRCC representative

Data Tools and Methods 19   (U.S.DOT 2020) who engages with the U.S.DOT TRCC. Each state has an established system for managing police-reported transportation crash and injury data, and most aim to conform to guidance offered in NHTSA’s MMUCC, which offers voluntary guidance, not mandates, for how to standardize data collection and reporting in relation to crashes. The current edition has been underway since 2016, which was prior to the establishment of e-scooter programs in the United States. Many states refer to the Association of Transportation Safety Information Professionals (ATSIP) publication, Manual on Classification of Motor Vehicle Traffic Crashes (ATSIP 2017), which offers voluntary standards for classifying and reporting traffic crash data. The current edition was published in 2017 and does not provide any micromobility standards, though the committee working on the next edition is considering this addition. In addition to MMUCC, NHTSA has equipped states with Model Performance Measures for State Traffic Records Systems (NHTSA 2011), which has not been updated since 2011 and does not currently incorporate e-scooter related traffic records guidance. While federal transportation legislation has established uniform procedures for state traffic safety information system funding (via data improvement grants that flow through SHSOs), there are no legislative requirements for states for how they structure or fund their safety information systems or how often they update their systems. Thus, opportunities to improve the consistency and reliability of police-reported crash data and other data needed to understand e-scooter safety issues will depend on the interest and involvement of TRCCs to advance the state-level data collection systems. TRCCs in each state may operate in unique ways and may be made up of different stakeholder groups with different safety data priorities. What they have in common is they all may serve as partners in providing updates to communities, offering forums for collaboration on projects and planned system improvements, recommendations to transportation agency leaderships, and support to local groups in the pursuit of data and training. While many state injury data systems remain far from reaching their goals of providing con- sistent and reliable traffic injury data related to e-scooters, there are clear examples of progress being made in terms of federal guidelines and state-level initiatives. For example, in summer 2022, NHTSA announced that for 2022 data, they will be incorporating new nonmotorist data elements in their Fatality Analysis Reporting System (FARS) and the Crash Report Sampling System (CRSS). Those manuals are used for coding and will be released with the public release of the 2022 data. Those data elements have included clarifying nonmotorist device types and motorization types. For riders of nonmotorized devices, the type of device now includes several micromobility vehicles classified by the SAE J3194 Standard, including standing or seated scooters. Similarly, clarifying codes for motorization type are included for devices operated by “nonmotorists.” This addition to FARS classification could assist in better data reporting for fatal crashes. Some states have begun to include scooter riders explicitly as a person type in police crash data. For instance, the Tennessee Integrated Traffic Analysis Network (TITAN) crash database added a “Pedestrian on Electric Scooter” and a “Person on Personal Conveyance” person type to its crash reporting form in May 2020. Though it was not part of ongoing TRCC data system plans, it was initiated by an individual within the Tennessee Department of Safety and Homeland Security and was implemented relatively quickly. Although there has not been any formal training to law enforcement agencies in the state about the change, inspection of the TITAN crash data shows that first responders were using it regularly at the beginning of 2021, and e-scooter crashes are now more easily identified in the datasets. For the main e-scooter market (Nashville), approxi- mately two-thirds of e-scooter crashes are now consistently coded with this person type. There are still “false positives”—people on other personal conveyances (e.g., electric skateboards, powered wheelchairs) are also coded as this person type—so further collaboration is needed between first responders, TRCC members, and crash data users to continue to enhance the data system.

20 E-Scooter Safety Toolbox It’s worth noting that even with reliable and consistent crash reporting standards, many exclude single-person events that don’t involve a motor vehicle, such as a bicyclist crash on a railroad, a pedestrian slip on a grate, or an e-scooter run off road crash. Many states have thresholds for reporting based on rather antiquated, vehicle-based criteria, such as motor vehicle damage thresholds, that often result in crashes being excluded from reporting, even if damage to a micromobility device is evident. While most states consider any motor vehicle crash that results in an injury to be reportable, injuries are not always obvious and therefore may not be reported. In addition, most state crash systems do not capture injuries, including fatal injuries, involving single vehicle e-scooter collisions since these vehicles are not typically classified as “motor” vehicles, despite being electrified. In cases where state crash data are lacking and/or systemically biased against gathering micromobility-related incidents, there are myriad examples offered in the BTSCRP Web-Only Document 5 on ways local micromobility program operators are collaborating with others to access additional data. For example, in Portland, the Portland Bureau of Transportation (PBOT) collaborated with the Multnomah County Health Department to evaluate e-scooter injuries using injury surveillance data collected by the health department. Many transportation agencies have established partnerships with metropolitan planning organizations (MPOs) to conduct routine travel surveys; these could be updated to build in standardized data items related to e-scooter safety, near hits, perceptions of risk, self-reported behaviors, and travel trends. Travel surveys that are culturally and linguistically inclusive and utilize robust sampling methods can help gather important information, including information on demographic issues and travel purposes, that may further contextualize safety concerns and perceptions. Timely Timeliness will differ across traffic crash and injury surveillance systems, with syndromic surveillance systems generally having more timely data (data available in near real time) than trauma registries, hospital/emergency department discharge datasets, police-reported crash datasets, and death registries (data available 1-2 years after the event). However, there may be a trade-off between timeliness and other domains of data quality, such as reliability. There are opportunities to improve the timeliness of data, particularly information that is gathered routinely from operators and/or travel surveys via partnership with MPOs or others conducting routine travel surveys. Additionally, partnerships with key data owners can help establish more timely access to data held by others. San Francisco’s Vision Zero program funded a position for an epidemiologist who was embedded in the transportation and transit department activities, but had access to the public health injury surveillance data. Some cities mandate timeliness of data provided by operators through their permitting processes, requiring incident reports within a certain number of hours or days as well as regular ridership updates or access to dynamic data systems. The city of Chicago requires licensees for shared e-scooter services to comply with Mobility Data Specification (MDS) for open data. They must also provide a public Application Programming Interface (API) to present the “location of charged, rentable, and available scooters” and this must adhere to the General Bikeshare Feed Specification (GBFS) standard. This enables access to real time vehicle availability for third-party vendors so that this information can be integrated with other shared services such as ride hailing and transit. Licensees are required to provide details of incidents, crashes, or police actions in quarterly reports, in addition to reporting each incident within 24 hours of the incident. Accessible Ideally, as data quality and reliability improve, e-scooter safety data and performance metrics can be made more accessible to the public and more readily shared to stakeholders and

Data Tools and Methods 21   decision-makers. In most examples seen to date, e-scooter safety data are made available in a pilot or annual program report. While most agencies collect additional data (via operator reports, complaint lines, health care or police reports), there are few robust examples currently in micro- mobility program development of e-scooter safety and injury data available through easily accessible and/or interactive interfaces. Moving forward, there are opportunities for agencies to enhance their data dashboards, such as those that may support Vision Zero programs (seen cities New York City (https://vzv.nyc/), San Francisco, and North Carolina (https://ncvisionzero .org/data-analytics/visualizations/)), with more information about e-scooter safety. In general, record-level (i.e., individual person level) injury surveillance data may not be accessible to the public due to Health Insurance Portability and Accountability Act (HIPAA; https://www.hhs.gov /hipaa/index.html) and the need to protect personal identifiable information among persons injured in e-scooter incidents. However, this does not preclude the sharing of aggregate informa- tion, such as tabular datasets or data visualizations of injuries. Additionally, agencies can share information about complaints (by e-scooter riders or other road users) as well as lists of safety projects, new parking or infrastructure, or other programs underway. Accessible data is an important tool for routine public engagement and for sharing safety messaging with the public. Denver’s current shared micromobility program requires public access to data. Operators must make the data available via a public API that meets the standards of GBFS, and that displays the data via a “real time dashboard” of e-scooter location and availability to riders and to the city. In addition, via the real time dashboard, operators must provide the city with data that includes information on incidents, complaints, and crashes (City and County of Denver 2020). See Figure 6 for a visual of the data interface. Integrated Safety data integration has been a challenge in the transportation field, with political and institutional barriers related to data ownership and funding; concerns related to data privacy and security; and technical issues related to data missingness and linkage methods. Despite these issues, there is widespread recognition of the value of data integration to help identify network-level risks and develop data-informed plans and programs. There are many examples of transportation agencies working to integrate crash data with roadway inventory data and other spatial data (such as land use) and trip generation data (such as school and health care features). This is an opportunity for local and regional agencies to leverage SHSO funding for safety data improvements, and to begin building broader coalitions with key data partners. The Open Mobility Foundation hosts the “Mobility Data Specification” (https://www.openmobilityfoundation.org /about-mds/) to support standardized data, communication, and data sharing across entities. Additionally, the National Association of City and Transportation Officials (NACTO) Guidelines for Regulating Shared Micromobility has a 2019 policy (https://nacto.org/wp-content/uploads /2019/05/NACTO_IMLA_Managing-Mobility-Data.pdf) on managing mobility data, including data sharing processes. These resources may be useful in prompting conversations around data sharing and data integration. The city of Chicago is building a foundation for more advanced data integration through partnership cultivation. The city has sought to collect additional data to evaluate program safety performance. In addition to vendor provided data, the city collected data from its 311 service and an online feedback tool developed to gather feedback from e-scooter users and non-users. Additionally, the city worked with a local university to develop a survey for shared e-scooter riders during the 2020 pilot and worked closely with Vision Zero team to review police-reported crash data. During the two pilots, the city worked with the Chicago Department of Public Health to assess injury data. Through the collaboration, they were able to implement a coding scheme for e-scooter injury records that allowed identification of e-scooter crashes in the emergency

Figure 6. Map showing e-scooter trips 2018–2022 in Denver, CO. Source: Ride Report (2022a).

Data Tools and Methods 23   department data. No update of this information has been published since the 2020 pilot report, likely due in part for public health departments to focus their attention on COVID-19 and its societal impacts. While the city has many steps to go before their evaluation program can rely on truly integrated data, these initial steps in the pilot program provide important building blocks for future enhancements. Performance/Outcome Oriented With the safety data needs and possibilities for improvement, it is important to tailor data goals to serve the specific needs of e-scooter safety management and program delivery. This means taking into consideration how more consistent, reliable, timely, accessible, and/or inte- grated data can be best applied to answer important questions or fill a gap in knowledge, and now new data initiatives can best complement or dovetail with other ongoing efforts or programs. While data are needed most by local agencies responsible for day-to-day e-scooter programs, they may be used in key state processes, such as to help monitor implementation of Strategic Highway Safety plans that include e-scooter elements. As mentioned earlier, data are an important tool for public engagement and increasing awareness of safety issues. The city of Denver requires comprehensive operator reporting on shared micromobility usage to adhere to the MDS. During its pilot program, Denver Department of Transportation and Infrastructure (DOTI) staff noted that managing data in different formats for each operator and mode significantly increased the complexity and effort required. However, staff deemed the public real time data and the historical usage data useful to identify problems, such as locations where e-scooters that might be parked or abandoned in the right-of-way, or where a geofence might need to be created (City of Denver 2021). Similarly, staff for the PBOT have been able to navigate through an assortment of data needs and options to begin to develop data systems that now clearly tie e-scooter ridership on sidewalks to roadway characteristics, such as the presence of a bike lane and roadway classification. This connection allows them to use ridership data to proactively identify bike network gaps and prioritize future bike network investments.

Next: Chapter 5 - Additional Resources »
E-Scooter Safety Toolbox Get This Book
×
 E-Scooter Safety Toolbox
Buy Paperback | $64.00
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Since their introduction in the United States in 2017, the use of electric scooters (e-scooters) has expanded to the streets and sidewalks of many cities, and all indicators point to continued growth.

BTSCRP Research Report 9: E-Scooter Safety Toolbox, from TRB's Behavioral Traffic Safety Cooperative Research Program, presents findings from a multiyear research effort that sought to build on existing research to date, identify key gaps in knowledge and data related to e-scooter behavioral safety, and develop evidence-based guidelines that can enhance the coordination of behavioral safety programs and countermeasures with a broader toolbox of approaches to improve safety for all road users.

Supplemental to the report are BTSCRP Web-Only Document 5: E-Scooter Safety: Issues and Solutions and a presentation.

READ FREE ONLINE

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!