Data in All Decisions: The Evolution of Data-Driven Safety Analysis

Where should we install roadway safety countermeasures? Will people be safer if we add that crosswalk? Which of the intersection designs we are considering is safest?

These are the kind of questions transportation practitioners ask themselves every day to try to make roadways safer but, with so many variables at play, it can be difficult to know which of many possible decisions is best.

That’s where data-driven safety analysis (DDSA) can help.

DDSA is an approach to roadway design that says, to the greatest extent possible, practitioners should incorporate the use of data in their decision making.

FHWA first promoted DDSA in 2015 as part of the Every Day Counts (EDC) program. The EDC DDSA team trained staff at agencies across the country on data-driven tools like the Highway Safety Manual (HSM) Part C, approaches like systemic analysis, and proven countermeasures like Local Road Safety Plans.

The power of these approaches and tools is they can help to not only identify locations of past severe crashes (“hot spots”) but also to predict where future crashes might take place, so agencies can prevent them. DDSA predictive tools have become popular across the country since 2015 because of the opportunity they provide to proactively save lives.

While the resources initially promoted during EDC are still valid, DDSA has evolved over the last 9 years. Tools and strategies have increased in number and quality. New data sources have become available. Agencies have more options for safety analysis than ever.

Additionally, as DDSA continues to evolve, researchers, public agencies, and industry are exploring the use of artificial intelligence and machine learning that may, in the future, allow them to identify improvements to predictive and systemic approaches.

In some cases, advances may be made in the statistical understanding of cause and effect with these newer analysis techniques. In other cases, these newer analysis techniques may improve the efficiency of working with available data and analysis of it at the scales needed to tackle the roadway safety crisis. Newer professions such as data scientists are now working in the transportation industry to help find new ways to solve problems with DDSA.

For agencies selecting safety countermeasures, these advances mean more to study, understand, and incorporate, which can overwhelm practitioners. However, according to FHWA’s DDSA team, it is crucial, and possible, for agencies to match these evolving DDSA approaches to their available resources.

Progress Over Perfection

A current misconception is that, because data and tools have improved, they can predict crashes perfectly and therefore only predictive models should be used in DDSA. When assessing which tools to use, FHWA DDSA Program Manager Matt Hinshaw says practitioners should not let perfect be the enemy of the good.

“The predictive tools like the HSM Part C and applying crash modification factors are useful, and people should use them more as we seek to further integrate safety into the project development process,” Hinshaw said. “But it is easy, especially for technically minded people, to only incorporate DDSA if we can predict the results of our decisions with 100-percent accuracy. If not, we don’t want to use it.”

Hinshaw says DDSA approaches and data sources have increased, but none is perfect, and that is okay.

“We have broadened our horizons as to which methods fall under DDSA. This can include any form of evidence-based results. Examples are predictive models, the systemic safety approach and risk-based approaches, surrogate measures of safety, design flag assessments, Safe System Approach frameworks such as intersection conflict point analysis, transportation system management and operations strategies, and video analytics.”

Another misconception Hinshaw has heard is that an agency does not have enough data for DDSA.

“While more data is usually better, it is not a deal breaker if you don’t have robust data on hand,” Hinshaw said. “DDSA has evolved so that practitioners can usually incorporate it at some level no matter their situation. The idea is to start where you are with what you have and improve over time. We can find a way to use data to improve safety for any agency, especially with the systemic safety approach.”

One Approach, Many Tools

The overarching concept for DDSA is that a standardized, consistent, data and evidence-driven approach will lead to design decisions that prioritize safety.

DDSA can also include qualitative assessments that support project needs and design elements, including Road Safety Audits and public and stakeholder surveys, as long as they are guided by DDSA principles. In addition, a variety of tools can be used during the same project, such as analyzing a proposed interchange with a variety of resulting metrics to compare.

“Think of DDSA like a menu of options, with different methodologies to choose from based on your available data and desired output,” said Jerry Roche, FHWA Safety Integration Team Leader. “Some methods may be preferred in certain contexts. Human judgment is still needed.”

Ideally, agencies will incorporate DDSA into every stage of the project development process, placing safety on equal footing with other project impacts like operations and using data to justify their decision making.

A Place for DDSA and DDSA in Its Place

The goal of DDSA is not to replace human decision making, but to augment and inform it. “Engineering judgment and reaching consensus on project teams made of professional safety, traffic operations and design staff should always be the bedrock of the decision-making process and help work through DDSA,” Hinshaw said. “We have great opportunity to save lives here, if we’ll take advantage of the data and tools we have.”