Introduction

I strongly believe that road (un)safety is an overlooked problem, which is why I started a new #DataVisualization project to present traffic crashes deaths in the United States. This project is an effort to create awareness among citizens and policymakers regarding road (un)safety. It is the capstone project for the Master of City and Regional Planning I am studying at University of California, Berkeley.

The COVID-19 pandemic generated the proliferation of official and independent data analyses, visualizations in social media (e.g., me), digital newspapers (e.g., New York Times, Financial Times), and webpages (e.g., Our World in Data, Coronavirus Tracker). The access to real-time information about the number of cases, deaths, and hospitalizations transcendent the sphere of governments’ public health decision making and became part of the regular information consumed by many on the internet and in the news. The availability of the data probably helped to raise awareness about the risks and danger of the pandemic in a day-to-day basis.

By the end of 2022, over one million people have died from COVID-19 in the United States (Ritchie, et al., 2020). In comparison, between 2020 and 2021, over 81,000 people died from traffic fatalities in the United States (National Center for Statistics and Analysis, 2022).

Approximately 40,000 people die each year from traffic crashes in the US.

This is significantly lower than the COVID-19 numbers but is a phenomenon that has been happening for years (see the following figure), way before COVID-19 hit, and it will probably continue without significant reductions in the following years or decades. Despite the above, these numbers rarely make the news or go viral (although specific events might). Traffic fatalities are far away of the numbers seen with COVID-19 in recent years, but the level of awareness about the risk of traffic deaths seems to be very low in comparison to the real danger that the statistics show.

Here, I explore and analyze data of traffic fatalities at a national (United States) scale to build appealing and suitable visualizations that address the considerable heterogeneity by mode, time of the day, day of the week, type of road, victim gender, victim age, and other available variables.

Inspired by the proliferation of data analyses with COVID-19 data, here I explore and analyze traffic crashes data in the United States to build visually appealing and suitable data visualizations. I am also sharing those visualizations on my Twitter account. Traffic crashes are not as disruptive as COVID-19 and data is also less dynamic and available, which represents a challenge to call the attention of a general audience and public opinion. My goal is to build a “tracker” of traffic crashes data that includes the overall framework and dimension of this issue, but also a dynamic and weekly update of the traffic emergency events. To address this, I am combining two sources of information managed by the National Highway Traffic Safety Administration (NHTSA):

  1. The Fatality Analysis Reporting System (FARS), a nationwide census of the fatal injuries suffered in traffic crashes. It contains detailed information on all traffic crashes that occur on public roads and that involve at least one vehicle that is involved in transport and in which at least one person dies within 30 days of the crash.
  2. The National Emergency Medical Services Information System (NEMSIS), a national system to collect, store and share emergency medical services data, updated weekly.

This is work in progress, but one of the takeaways of my first analyses is that there is a lot of data that has been gathered for decades and is usually under analyzed. There are availability and access challenges that I have experienced as a user of the information, such as the process to access and download the databases, and that I want to discuss further in this work.There are other data sources that are relevant to this issue and that I could potentially integrate to this project, such as the Environmental Protection Agency (EPA) data on the type, size, weight, power, and other characteristics of the vehicles sold in the US and that are a crucial factor to explain some of the trends seen, such as the decrease in vehicles’ occupants injury severity and increase in the number of fatalities among the most vulnerable road users.

FARS

Fatality Analysis Reporting System

Crashes

From the Fatality Analysis Reporting System Analytical User’s Manual, 1975-2020:

  • Accident – (1975-current): This data file contains information about crash characteristics and environmental conditions at the time of the crash. There is one record per crash.

The number of traffic deaths today are similar to the numbers from 1990.

Drunk driving

The percentage of fatal crashes with drinking drivers involved has been gradually decreasing.

Light condition

Fatal traffic crashes occur at every lighting condition…

Type of road

… at every type of route signing…

Day of the week

… all days of the week…

Time of the day

… and at every time of the day.

Persons

From the Fatality Analysis Reporting System Analytical User’s Manual, 1975-2020:

  • Person – (1975-current): This data file contains information describing all people involved in the crash including motorists (i.e., drivers and passengers of motor vehicles in transport) and non-motorists (e.g., pedestrians, pedalcyclists, and occupants of motor vehicles not in transport). It provides information such as age, sex, vehicle occupant restraint use, and injury severity. There is one record per person.

Injury severity

Almost half of all the persons involved in fatal crashes have fatal injuries.

Mode - Person type

As shown in the following figure, the number of pedestrians and cyclists involved in fatal crashes has been increasing since 2010 and are also the road users with the highest risk of being killed (over 90% of pedestrians and cyclists involved in fatal crashes are killed, compared to around 50% of drivers and 30% of other vehicle occupants).

Age

Younger people are dying less in traffic crashes than 30 years ago (across all modes), but the number of traffic deaths among 55 and older has been increasing, especially among pedestrians.

Pedestrians and Bicycles

From the Fatality Analysis Reporting System Analytical User’s Manual, 1975-2020:

  • Pbtype – (2014-current): This data file contains information about crashes between motor vehicles and pedestrians, people on personal conveyances, and bicyclists. Data from the crash are entered into the Pedestrian and Bicycle Crash Analysis Tool (PBCAT). The output fields from PBCAT, including the pre-crash actions of the parties involved (crash type), are included in this data file. There is one record for each pedestrian, bicyclist or person on a personal conveyance.

The number of pedestrians and cyclists involved in fatal crashes increased from around 6,000 in 2014 to 8,000 in 2020.

Crosswalks, Sidewalks, and School Zones

Pedestrian and cyclists’ fatal crashes occur mostly in places where there are no sidewalks, no marked crosswalks, and no school zones. Nevertheless, over 30% of fatal crashes occur in places where sidewalks are present and over 15% where marked crosswalks are present.

Location

About 35% of fatal crashes involving cyclists and 25% involving pedestrians occur at intersections or intersection-related locations.

Motorist Manuever at Intersections

Approximately 1,700 fatal crashes involving pedestrians occurred in intersections in 2020 (out of 7,000 in all locations). In the vast majority of those crashes the driver was going straight through, but a significant number of drivers were doing left or right turns.

Vehicles

From the Fatality Analysis Reporting System Analytical User’s Manual, 1975-2020:

  • Vehicle – (1975-current): This data file contains information describing the motor vehicles in transport and the drivers of motor vehicles in transport who are involved in the crash. There is one record per motor vehicle in transport. Parked and working vehicle information is in the Parkwork data file.

Speeding

The following graph shows the number of vehicles involved in fatal crashes that were reported as speeding or not speeding and the travel speed at which they were travelling before the crash from 2009 to 2020. It is surprising that around 25,000 vehicles per year that were reported as not speeding were traveling at over 91 mph (according to the same database) when no state has a speed limit over 85 mph. This raises questions regarding the definition of speeding, the way in which this is being determined, and the method applied to report the crashes and the consequent data. There are reliability and data reporting challenges that are crucial to improve our understanding of traffic deaths and advance in effective solutions to resolve it.

Vehicle occupants deaths

## 
##      0      1      2      3      4      5      6      7      8      9     10 
## 728324 856223  63084   9046   2241    571    154     55     18     14      5 
##     11     12     13     14     15     17     18     22     23 
##      5      1      4      2      3      1      1      1      1

Driver’s license and history

NEMSIS

National Emergency Medical Services Information System

Traffic/Transportation Incident

Instructions to access the NEMSIS data

https://nemsis.org/

View Reports -> EMS Data Cube -> Connect to the EMS Data Cube -> Login

Username: nemsispublic Password:

Design -> OLAP Report -> Connect to Cube

  • Summary Measures: Count of Events

  • Rows: Day (from Year - Quarter - Month - Day)

  • Columns: Vehicular - Pedestrian or Other Injury Factor

  • Filters: Compliant Reported by Dispatch: Traffic/Transportation Incident Vehicular

  • Filters: Year: 2023 (from Year - Quarter - Month - Day)

Traffic Incidents

Compliant Reported by Dispatch: Traffic/Transportation Incident

## Warning: Removed 2 row(s) containing missing values (geom_path).
## Warning: Removed 2 rows containing missing values (geom_point).

Auto Running Over or Impacting Pedestrians or Cyclists

Traffic Incidents -> Auto v. Pedestrian/Bicyclist Thrown, Run Over, or > 20 MPH Impact

## Warning: Removed 2 row(s) containing missing values (geom_path).
## Warning: Removed 2 rows containing missing values (geom_point).

Notes

Next steps:

  • Discussion about the data available, who and to what extend is used and useful (relate with CSCRS newsletter submission)

Graph feedback:

  • Model year with non-occupant fatalities

  • Fleet composition by year (EPA sources)

  • Clean Air Act