Rural roads can change considerably from one part of the week to another. Traffic patterns, travel purposes, road conditions and driver behavior may differ between weekdays and weekends, creating different circumstances when a crash occurs.

A 2026 study in Transportation Engineering examined these differences using a combination of machine-learning techniques and statistical crash models. The research focused on light-vehicle crashes and explored why similar road environments can produce different levels of injury depending on when a crash happens.

Looking Beyond Basic Crash Factors

Crash-severity research commonly examines factors such as road design, vehicle characteristics, driver information, lighting and collision type. However, treating each factor separately can miss situations in which two conditions influence one another.

For example, a particular road feature may have a different relationship with crash severity on a weekday than during a weekend. The researchers therefore incorporated time-of-week interactions into their analysis rather than treating weekdays and weekends as simple calendar categories.

The study examined a divided two-way rural corridor of approximately 175 kilometers through mountainous terrain. Crash records from 2017 through 2022 were analyzed, with incomplete records removed. The final dataset contained 2,199 light-vehicle crashes. About 77% involved property damage only, while roughly 23% involved injury or fatal outcomes.

Combining Statistics With Machine Learning

One of the study's central features was its hybrid analytical framework. Instead of relying on a single statistical model, the researchers combined econometric methods with machine-learning-based segmentation.

The approach attempted to identify groups of crashes that shared hidden characteristics. Two types of discrete-choice models were assessed within the framework: a random-parameters logit model and a generalized ordered logit model. The researchers also used an association-rule-based feature-selection method to identify important interactions involving time of week.

This process produced six notable time-of-week interaction variables involving factors such as driver licensing, lighting, collision type, vehicle age, lane width and road-surface condition. The final analysis divided crashes into three broad severity-related segments, helping reveal patterns that could be obscured in a single combined model.

Weekday Crashes Show Different Patterns

The findings indicated that weekday crashes could fall into distinct groups with different risk profiles. One lower-severity group was more closely associated with crashes occurring on or near the shoulder.

A separate critical group showed stronger connections with mountainous road conditions and older drivers. In particular, drivers aged 60 and above were more prominent in the higher-severity weekday segment. The researchers suggested that mountainous terrain may create additional challenges because of road geometry and restricted visibility. The findings also support attention to roadway shoulders, especially because shoulder availability was associated with lower injury severity in the analyzed data.

Weekend Crashes Stand Out

The weekend pattern was particularly notable. The model identified a distinct segment consisting exclusively of weekend crashes, with this group associated with relatively high severity. Two crash characteristics were especially prominent: rollovers and roadside collisions. Roadside crashes were nearly twice as likely in the weekend high-severity segment compared with the relevant weekday comparison. Rollover-related outcomes also showed a substantially higher likelihood in this group.

What the Findings Mean for Road Safety

The study points toward a more targeted approach to rural-road safety. Instead of applying identical measures throughout the week, safety planning could account for the circumstances associated with different time periods. For weekday conditions, the research highlights the potential importance of wider shoulders, particularly where lower-severity crashes are concentrated.

Why Timing Matters

The broader lesson is that crash severity cannot always be explained by looking at road, vehicle or driver characteristics independently. The same factor can behave differently when combined with another condition.

By separating hidden crash patterns and examining interactions involving time of week, the study provides a more detailed picture of rural-road safety. At the same time, the researchers acknowledge limitations, including the use of police-reported crash data and the absence of some detailed roadway information.

The research shows that the timing of a rural-road crash can be an important part of understanding its severity. Weekday crashes displayed different characteristics from weekend events, while mountainous terrain, older drivers, roadside locations and rollover events emerged within specific high-risk segments.

Rather than treating every crash as part of one uniform population, the hybrid modeling approach reveals several distinct patterns. Such information can help road-safety planners consider when and where particular infrastructure improvements may have the greatest relevance, while also demonstrating the value of combining statistical modeling with data-driven analysis.