SPR 20-598: "Vehicle on Shoulder" and Crash --- Correlation or Causation

The Kentucky Transportation Cabinet (KYTC) has been partnering with Waze through the Connected Citizens Program to access traffic information reported by Waze users, including accidents, traffic jams, potholes, objects on the road, vehicles stopped on shoulder, and other hazards. Such information has been used by KYTC to provide situational awareness to the travelers and support real-time incident management. After reviewing traffic data alongside these crowdsourced reports, the ITS Branch at KYTC has observed frequent chain of events that appear to be in the sequence of "Vehicles On Shoulder", speed drops to well below free-flow condition, and then reports of "Accident". A better understanding of the correlation and/ or causation among these alerts can help Kentucky develop operational strategies and policies to prevent future crashes and fatalities. This study aims to answer the following research questions: (1) Does the aforementioned chain of events demonstrate causation between alerts such as "vehicle on shoulder" and "accident"? (2) Is a particular type of Waze alert, in conjunction with other conditions such as slowdown, more likely to lead to a crash? (3) Explore the nature of the crowd-sourced alerts. ( 4) What is the temporal/ spatial pattern of such chain of events? When and where does it occur most frequently? What roadway and traffic conditions are more likely to be associated with severe crashes? (5) Is there a pattern on the time frame between initial "vehicle on shoulder" alert and the subsequent "accident" alert?