Who are we?
Real Safe Cars was created by Ben Shiller and Bhoomija Ranjan in 2019 after they realized that existing public car safety information presented an incomplete picture of car safety. The ranking process was inspired by their joint paper with Siqi Liu, 'Optimal Presentation of Quality Ratings: Application to Coarsened Automobile Crashworthiness Ratings'. Shiller and Ranjan are both economists at Brandeis University who specialize in structural modeling.
Why rank cars?
One founder went through a crisis. A family member was very sick, and a friend died in a car accident. Through this difficult period, he developed a deep desire to save lives, but it was not immediately obvious how he could --- he was not a doctor or chemist, but rather a professor of economics/statistics. In the process searching for a new car – he had his eureka moment. He could develop better safety rankings. Existing car safety ratings were opaque, and moreover it was difficult to distinguish between the many cars rated as “safe.” He believed he could do so much better. In truth, it might have always been his destiny --- his family has long had a passion for car safety, always choosing the safest-rated white-colored cars, which are easiest for others to see (well, except in a blizzard). At family gatherings, there are a comical number of white cars.
He sought out the best colleagues for the project. His two like-minded cofounders were equally enthralled and enthusiastic about the idea. Together, we developed this idea from concept to reality.
What do we do?
RealSafeCars presents car safety ratings that were developed by a process of computer modeling. Our models incorporate a range of different data sources. These ratings are presented in a meaningful fashion so that the magnitude of the differences between different models are clear.
Why do we believe our ratings are better?
Car safety ratings organizations do necessary work and saved countless lives. Our ratings are presented as full alternatives to those ratings. Rather, we believe that those ratings should be used as a starting point. By combining those ratings with accident statistics and using computer modeling to assess what car safety tests really show us, we believe our ratings offer some big advantages.
- Others’ ratings are based on very particular testing conditions. Others’ ratings will, for example, provide good information about how well someone will fare if impacted from the side by a 3000 pound vehicle traveling at 30 miles per hour. But what happens if impacted at 50 miles per hour, or by a heavier vehicle? Their ratings are not well-designed to answer this question. We designed our ratings to better evaluate crashworthiness under real-world crash scenarios.
- Others’ ratings often award the top rating to many cars. But some cars with the same discrete rating are much safer than others. Only our ratings provide information about which cars are really the safest.
- Others’ ratings are not comparable across different model years. Our ratings are.
Why are causality analyses important?
We strive to predict which cars protect occupants best. But proper analysis requires great care. Raw correlations between features and fatality rates may lie. Consider minivans. Is their low fatality rate attributable to safety features, or habits of their drivers? Maybe minivan drivers transporting their children drive unusually cautiously, explaining the low fatality rate. Our readers, however, want to know how well a car will protect them, not the other way around. By combining artificial intelligence methods --- which do not account for causality --- with advanced causality analyses, we are able to answer the question our readers truly care about – how well does a car protect its occupants?
How do I find the model year of a vehicle?
All vehicles sold in the US must have a unique 17-character Vehicle Identification Number (VIN). The VIN is shown on the lower portion of the windshield on the driver’s side, just above the hood/windshield wiper. It is easily visible when looking through the windshield from outside the vehicle. The VIN should also be shown on the manufacturer’s label affixed to the door jam on the driver’s side.
The 10th character of the VIN denotes the model year, via the following coding.
VIN’s 10th character | Model year |
A | 2010 |
B | 2011 |
C | 2012 |
D | 2013 |
E | 2014 |
F | 2015 |
G | 2016 |
H | 2017 |
J | 2018 |
K | 2019 |
L | 2020 |
M | 2021 |