When I was about 5 years old, my family moved from Chicago to San Francisco. Though we could have flown, my parents made the unilateral decision to drive the 2,000+ miles. As a result, my mom, dad, brother, grandma and I embarked on a 7-day tour of middle America’s motels and drove for hours and hours and hours on end.
The author preparing the car for the journey ahead.
By some sheer coincidence, right around the time that I was on this journey (1995), the government commissioned a survey which consisted of asking drivers “what car trips did you take in the last 24 hours?” Respondents were asked to answer to the best of their recollection.
As a result, we have a collection of over 100,000 car trips Americans took in 1995. With this unique dataset, it’s possible to answer the questions of my young, existentially challenged – or more likely road-trip bored – mind: Who are all these people on the road, where are they going, and who are they traveling with? Here’s what I learned:
When are people driving?
First, I looked at the full dataset and, after removing absurd outliers, analyzed what time of day people were setting off.
If you look at the nighttime hours between 7 p.m. and 6 a.m., (i.e hours when the average person is driving for a purpose besides commuting to work), this is a beautifully even graph. Then, as the world wakes up, things start to get a little more interesting. Around 6 a.m, we start to see the first inklings of human indecision: How many times can I press the snooze button? Will I have breakfast on the road today or at home? What’s the latest I can leave and still make it to work on time?
These small variables affect when someone sets off for the day and gives the graph a small, but smooth, curve. Then when you hit 7am you see a significant spike and then, a little while after, a sharp drop. This may look unnatural but is easily explained: Commuters.
The peak at 7 a.m. illustrates all the people heading to work. Then the trough that occurs from 8 a.m.-noon reflects workers at their desks, unmoving. Of course, there’s another peak representing those who hop in their car to grab lunch around 1 p.m. and another spike when people head back out to pick up their kids or drive home at the end of the day 3 p.m. to 6 p.m.
One interesting component of this dataset is the fact that respondents were asked to record how long each trip took. With this in mind we can easily find the arrival times of each of the trips. This sharpens the first peak (which is consistent with people needing to arrive at the office by 8am) whereas it has smoothed out the lunch peaks and the commute home peaks. This can be explained by the arrival time at work being based on the company’s clock, and the departure being up to the individual.
How far are people driving?
Next, I wanted to see how far everyone was driving each time they got behind the wheel. The general trend is that shorter journeys are much more common than longer ones — and there is a smooth decline.
There were a few of noteworthy stats from this analysis. First, the 0-1 mile trip subset is much lower than the 1-2 mile group. One explanation for this finding is that if you’re traveling less than a mile, you may opt to travel on foot, as getting into your car proves to be a greater hassle than it’s worth. Another factor that may explain this could be the way the data was collected.
If you are accosted in the street by a pollster and asked to share all your car trips in the past day, and you were polite enough to actually list trips out one by one, you may not bother to mention the super short trips. You may consider the time you took the car to the park with the dog unnecessary to share, or you may have even forgotten about it because it was so insignificant.
This may seem far-fetched, but previous studies support the fact that respondents may not give exact answers if there is an easier, prettier, response to give. In the chart above, you generally see a decline in the number of journeys at longer and longer distances, however, there are a few uncharacteristic peaks at 5 mile intervals. This is due to an effect known as “statistical heaping.” Statistical heaping occurs when survey participants are unsure about an answer so they go with an aesthetically pleasing number such as 45 (rather than 46). This suggests that in many cases, round numbers will have a peak. So when a person is asked to quickly estimate something they don’t know perfectly they will cause this heaping. A good example of this is when nurses in Tanzania were asked to weigh newborn babies, and you see these peaks.
This lends weight to the theory that there potentially are a significant amount trips that measure a mile or less, but they weren’t noted due to the specific methodology for collecting this data.
How many passengers are in the car?
Another interesting finding from our dataset is the number of passengers in a car at any given time. Respondents were asked how many people were in the car with them for each trip, which allows us an understanding of how full cars tend to be depending on the time of day.
The general trend reveals that the average car has 1.8 people in it regardless of time of day. However, there are two notable departures from this average. First, the period of time from 3-6 am has been dubbed the “anti-social hours” because there tend to be far fewer passengers in cars (with an average as low as 1.3). The opposite of this occurs in the early evening when people are doing things with their friends and family. You can see there is a peak that hits approximately 2.2 people in a car on average.
Families going out for meals seems like a reasonable explanation for the passenger peak that occurs in the late day, but the large dip in the anti-social hours seemed less obvious. To better understand what was going on here, I looked at what length of journeys were being taken at different times. Findings below:
I discovered that car trips in the early morning with just the driver in the car tended to be longer journeys. This, I suspect, results from people who travel long distances for work — predominantly truck drivers and couriers. Both of these jobs require a single individual to travel very long distances, typically leaving way before dawn. Because so few car trips typically start at this time of day, it is reasonable to assume that the sample of trips in that time period is largely due to these professions.
So, if anyone ever invents a time machine, I could zap myself back to 1995 at a motel somewhere in Utah and tell a much younger me a few things.
(1) Drivers typically prefer to arrive at their destination for the day at 9 a.m. and want to leave at 5 p.m.
(2) Some people travel alone for long distances in the early morning but plenty more people travel with their families in the evening for short distances
(3) Americans are much more likely to talk about their 25-mile journey than their 24-mile one, and probably won’t bother to mention any trips that are a mile or less.
As much as I’d like to believe that showing this article to a 5-year-old me would be enough to relieve the boredom of a 7-day road trip, I’d guess that he’d be more interested in playing Angry Birds on my iPad.