A week ago my wife and I were driving a rented Toyota Corolla from Pittsburgh, PA to Cleveland, OH. Since my wife was tired and slept, I was able to run a series of experiments that she would never had let me get away with, if she had been awake. (She hates it when I drive 55 mph)

The Corolla had a gas mileage gauge that provided both instantaneous and average gas mileage read outs. The instantaneous reading was useless, since the numbers kept oscillating widely. However, the average read out was more useful. It allowed me to reset the gauge, then I could drive for 10 minutes at a given speed, and then see what the gas mileage was. Since there are many variables, I attempted to keep as many things constant as possible for a given experiment. I also repeated each experiment three times. Specifically, I reset the meter and drove for 10 minutes under the following conditions:

- 55 mph
- 70 mph
- 70 mph (drafting behind a truck)

Each time I got very similar results:

- 55 mph --> 53 mpg
- 70 mph --> 37 mpg
- 70 mph (drafting behind a truck) --> 43 mpg

Of course we have all been told this before, but running the experiment myself and seeing the huge difference made an impact on me. Driving 55 mph, instead of 70 mph, decreased fuel consumption by a whopping 30%! On the trip from Pittsburgh to Cleveland (and back) 250 miles, I could use 4.7 gallons, instead of 6.7 gallons of fuel.

As a result, I started to think about whether I really needed to get places as quickly as I usually think I do. There are plenty of choices we all make. Lindsay Leeven in his latest blog post posits that a 50% reduction in vehicle weight reduces the fuel consumption by 30%. I already drive reasonably small cars, so I am not ready to go out and drive a much lighter car, when safety is at the top of my list.

I began to think what it would take for drivers in the US to make a noticeable difference in the number of gallons of gasoline that are sold for use in automobiles and light trucks and vans. Suppose 30% of the drivers on the road changed their driving habits for one week, is there a statistic that would show the results?

I discovered that one can download a spreadsheet from the US government, which summarizes the daily average of fuel sold for use in automobiles. The data is provided on a weekly basis, starting in 1990. For kicks I graphed the data for every year, since 1990, bracketing the week of every year that includes Christmas: For example, in 2010 the US statistics were published for the week ending Friday December 24 2010, Friday December 31 2010, and Friday January 6 2011. What I noticed over the 20 year period is that the amount of fuel used during the week that includes Christmas has the following characteristics when it is compared to the weeks that precedes and follows Christmas:

- Week of Christmas never differs by more than 11.11% from the other two weeks.
- The average difference between the Christmas week and the other two weeks is 3.27%
- The standard deviation between the Christmas week and the other two weeks is 2.82%

Figure 1 illustrates the data. The red bars represent the weeks that include Christmas.

Figure 1: The red bars represent the weeks that include Christmas.

So, if 30% of US drivers decided to slow down for the week that includes Christmas during 2011 (* December 25, 2011 --> December 31, 2011*), and the only difference in their driving habits was that they slowed down to 55 mph on the highway, then I believe that a statically significant observation could be made that showed when people in the US set their minds to making a change, they can do so. Wouldn't that be nice if people in the US decided to make a point rather than be legislated to slow down! Everyone knows that legislation usually doesn't work.