A new academic paper looks for learnings in data gathered via collisiontracker.ie

Outside the Spatial Dynamics Lab on the northwest-side of University College Dublin’s (UCD’s) campus, there’s bicycle parking, surrounded by grass. 

Inside the wooden-paneled building, postdoctoral Vinayak Malaghan sits at the head of the table in the conference room with a PowerPoint open with his new research findings on hazards for Dublin’s cyclists.

When he started as a postdoctoral research fellow at UCD in December 2023, he said, his focus was on sustainable mobility. Malaghan would usually bike to work and to get around the city too, he says. His lab focuses on active travel. 

“People are focusing more on, okay, we should bicycle. We should increase the cycling activity,” said Malaghan.  

“Yeah, that's fine, but the other side would be the issue of safety. Which is not emphasised,” he said. “So I thought, like, okay, we have the data, and, let's try this.”

He started to use the data gathered by via collisiontracker.ie – a map-based platform co-founded by Dublin Inquirer, which crowd-sources first-hand accounts of hazards, near misses, and collisions. He also used BikeMaps.Org.

And, he started analysing reports to map the hazards facing cyclists.

Malaghan was the lead author on a paper published in the academic journal Transportation Research.

The analysis found that issues cluster in the city centre, around tram tracks, roundabouts, blocked cycle lanes, close passes, left-turns, and heavy vehicles.

An even bigger dataset would help him to better understand, Malaghan said. 

Before he had crunched the numbers, though, one night in April, Malaghan was cycling home from a conference and he was hit by a car, he says.  

He needed surgery. He ended up needing to go back to India for the remainder of the project.

Recovery

Although Malaghan said the car wasn’t going fast, he had broken three bones in his foot. 

“I went to St Vincent University Hospital. And they said, it's quite a nasty crack, a nasty injury. And then they said that we would do your surgery the very next day,” he says. “But I waited for two, three days.”

He was then told he couldn’t be booked for weeks, says Malaghan. A doctor in India warned him waiting could complicate the procedure, he says.

“It was a bad time,” he said. So Malaghan flew home, and got the surgery right away.

After four months, he started to walk slowly on his heel, he says. “And then started placing the ball of the foot gradually. And now it's been one year, I started, I resumed all my activities.”

Malaghan said he finally feels close to full recovery though there’s lingering symptoms.  “As soon as I get up in the morning, I still feel stiffness. I need to do some sort of stretching, then I feel more comfortable.”

His collision slots into his research categories: it was due to driver behavior, and a car turned right in front of him, crossing the cycle lane. Since Malaghan was going straight on the cycle path, he had the right of way.

The experience, he says, sharpened his interest in his research.

The study

Most official crash data – Garda reports, hospital records – miss near misses. 

But Brian Rogers, a research scientist at the Spatial Dynamics Lab, who also built and manages collisiontracker.ie, and was another author of the Transportation Research paper, says these close calls are critical to capture.

“Things happen higher up that severity pyramid because we weren't able to act on near misses,” said Rogers. 

“For every tragic accident, there's a much larger number of serious accidents, typically. Maybe it's at, like, a dangerous spot at an intersection, or it's a roundabout that has been poorly designed,” he says.

Irish Cycling Campaign member Kevin Jennings, who is also a lecturer in mathematics at the University of Galway, also said this is a really important aspect of this study. 

Recording near misses makes it possible to see where accidents might happen in the future, he said. 

Stacking together the written accounts people give when submitting accident reports on collisiontracker.ie allows for analysis too, says Rogers. 

Basically, “looking at patterns in the words that people used to try to establish themes about what might be going on in that area”.

The data

Once Malaghan had parsed and grouped the data, patterns emerged. 

The most dangerous time was between 7am and 9am – and incidents peak again around the after-work rush hour. 

The data logged said that drivers are responsible the majority of the time – with complaints that they had cut off cyclists, were speeding, or illegally parked.

“Close passing events by drivers are often a consequence of aggression, impatience, and risky overtaking,” the study says. 

And, “These behaviours are exacerbated in shared spaces and on narrow streets where dedicated cycling lanes are absent.”

The cyclist errors flagged in the data included red-light running and wrong-way riding. Infrastructure flaws – abrupt lane endings, and poor surfaces – were also highlighted.

Self-reporting may downplay cyclists’ faults, says Malaghan. 

“Possibly there is a bias that cyclists are favoring themselves. Even though the fault is from their side, they are not agreeing with that,” he says.

But they are anyway trying to understand in a broader way what causes a crash, he says. 

“So, for instance, you can say close pass. Close pass is one thing,” he said. “So we don't know who – we are not saying that driver is at fault. We are just saying the reason for the cause of crash.”

The data analysis also found that, usually, there’s no injuries. But the most common injuries were to the shoulder and the elbow.

More needed

The data isn’t perfect, for sure, Malaghan said.

Right now, incidents peak on Wednesday and peter off into the weekend. 

Malaghan said he can deduce why there were fewer incidents on weekends – because of commuting. But “there is no proper reason why it's getting high on the Wednesday or Thursday”.

It's based on 1,000 entries. “So if you have 30,000 or 80,000 sets, then possibly this distribution will be different,” he said.

Demographic gaps exist too. The tracker doesn’t record gender or background, which both Malaghan and Kevin Jennings, the cycling advocate and mathematician, limits what researchers can learn. 

Still, Jennings said he can see that the submissions are skewed towards a certain demographic. 

It’s largely professionals contributing, he says. “Because, you know, the data is all in the morning when man is going to work, and in the evening, when man is coming home from work.” 

“As opposed to Deliveroo drivers who'd be on the road all the time constantly seeing these incidents,” he says.

“Don’t get me wrong, at peak times, you're going to have more interactions between bikes and cars, so there's more incidents,” he said. “But it's very likely that the people contributing to Dublin Inquirer are niche.” 

“And probably, just from my experience in cyclic campaigning, they're probably middle-aged professional men, you know, white and white Irishmen, you know, but it's possible this demographic is over represented,” he says.

Busy roads

They’re also mapping the “why” onto the “where”.

For Malaghan, the most fascinating finding was the visual imagery of the hotspots. 

Danger spots concentrate at the heart of the city centre, following the main roads, and tapering out on the outskirts, he said –  like a sun.

“The regional roads are radiating,” he said. Then “they encircle these regional roads”.

Rogers says that’s down to frequency of travel. “Volume of active travelers, or volume of cyclists passing through that area.”

So sometimes, the mapping doesn’t identify the most dangerous spots, but rather where people experience danger most frequently, he says.

Consider a lethal cycling location that isn’t in a busy area. It “gets passed through, you know, once a month, it's not going to appear as a cluster of reports”, Rogers said.

Jennings, the cycling advocate, said the collision tracker came up at a council meeting in Galway, and the road engineers said they don’t use it because it’s not an official source.   

He wants to see this data collection more integrated with official bodies, he says.

Rogers said he’s open to this too. “I would love to see the project be adopted by anybody that could make it a little bit more universal.” 

Rogers said the bureaucracy in agencies can prevent action on this info.

“They would probably say, we don't act in sporadic once-off fixes. We act and think about Ireland as a network,” Rogers said.

“And we propose and fund schemes that improve entire networks. So there is no SWAT team or like, you know, Scooby Doo van that can drive to a dangerous location and fix it.”

Malaghan’s study avoids formal recommendations because it’s outside the scope of the data, he said.

“These things can be avoided”

Jennings said he wants to see data like that collected via collisiontracker.ie considered, as it flags the risks before someone has to suffer the consequences.

He said that even when something does happen, often the response isn’t enough.

“A girl was killed in Galway there,” Jennings said. “She was cycling on the footpath, and, you know, I’ve a girl, the same age. They go to the same birthday parties. You know, this is just the worst.”

“This stuff happens, and then nothing happens because you can't, you can't talk about that. You're using somebody else's grief,” he said. “But this risk is real.” 

“It's an accident,” he said. “But these things can be avoided.”

Funded by the Local Democracy Reporting Scheme.