In 1960, 12.1 percent of American workers went to work by transit, which was then largely privately owned. Despite (or because of) public takeover of almost every transit system in the country, transit’s share steadily declined to 4.7 percent in 2000. Then, in 2010, it crept up to 4.9 percent. The 2014 American Community Survey found that it has increased still further to 5.2 percent.
Since 2000, the increase in transit’s share has come at the expense of carpooling, which fell from 12.6 percent to 9.2 percent in 2014. Biking and walking also fell slightly from 3.4 to 3.3 percent. Driving alone, however, grew from 73.2 to 76.5 percent. So the increase in transit’s share did not translate to a reduction in the number of cars on the road. Indeed, using census carpool data and assuming that “5- or 6-person carpools” have an average of 5.5 people and “7-or-more-person carpools” have 7 people, there were 104.2 million cars commuting to work in 2000, 110.8 million in 2010, and 117.6 million in 2014.
One intriguing table (B08141) shows commuting data by the number of cars in the household. Nationally, about 4.5 percent of workers live in households with no cars. Of these, about 41.5 percent took transit to work, 20.4 percent drove alone, and 11.3 percent carpooled.
Here are some useful spreadsheets:
- Commuting data by urban area, table B08301;
- Same table in raw format, which includes margins of error;
- Commuting data by urban area by vehicles available, table B08141;
- Same table in raw format, which includes margins of error.
The numbers for the nation at the bottom of the first and third spreadsheets include all commuters, not just those in urban areas.
As reported by the Census Bureau, margins of error are at the 90 percent confidence interval. In other words, if there are 100,000 commuters in a particular category and the margin of error is 1,000, then it is 90 percent likely that the actual number of commuters is between 99,000 and 101,000.
To simplify the tables I use, I deleted the margins of error. However, they are important when looking at smaller urban areas or less-used forms of commuting. The margin of error for, say, the number of people who drive to work nationwide is around a tenth of a percent. But the margin for some small category, such as the number of people who live in households with no vehicles and walk to work in Akron, Ohio is more than 50 percent. Any category that reports fewer than about 300 people has a margin of error close to if not greater than 100 percent.
This can lead to misleading results. The 2013 American Community Survey reported that about 5,700 workers in the Albuquerque urban area lived in households with no cars. This grew to 7,900 in 2014, which seems a large jump. Even more suspiciously, the share who drove alone to work shrank from 39.9 percent in 2013 to just 19.6 percent in 2014. I suspect this drastic change is more due to sampling error. Fortunately, the Census Bureau also reports five-year averages, which draw upon five times as much data and therefore are more reliable.