This is the post where I bore you with graphs, but its not too bad. As a followup to last friday’s <a href=”https://headwatersolver.wordpress.com/2014/01/31/is-transit-just-compressed-traffic/”>post</a> about the correlation (relationship?) between transit usage and commute times, I promised I’d look at some other things that might relate to transit, traffic and commutes. I also included Atlanta, at the request of a Facebook friend. The results of that search led me to some follow on questions, which I’ll address in a later post.
In addition to the received “commute time” and “transit share of commute”, I also found “congestion index” from the <a href=”http://www.rita.dot.gov/bts/sites/rita.dot.gov.bts/files/publications/national_transportation_statistics/html/table_01_71.html”>Bureau</a> of Transportation Statistics. It is a controversial measure, as congestion is a subjective and experiential variable. I you plan for congestion and leave a half hour earlier every day so you can unwind, you have a much better experience with congestion than a tourist who needs to make a flight out of town in rush hour. For bikers and walker, traffic congestion is a plus, not a minus, as traffic’s relative speed and absolute danger is much less than during free-flow times of day. However, the congestion index has been collected consistently for over a decade and is available for all metropolitan areas. I am humble that data availability trumps data quality, every time.
I also collected GDP by metro in 2004, 2006 and 2008 from the <a href=”http://www.bea.gov/newsreleases/regional/gdp_metro/2008/gdp_metro0908.htm”>Bureau</a> of Economic Analysis. GDP has its controversies as well, but they are not nearly as persuasive as the controversies over Congestion index. GDP measures all transaction, but we could all agree that a $140,000 house purchase contributes more to the economy than a week’s stay in the Intensive Care Unit.
Finally, I collected city, metropolitan and population change rates (birth, death, immigration, and emigration rates from the <a href=”https://www.census.gov/popest/data/historical/2000s/vintage_2006/metro.html”>census</a> for 2006. God bless them for having all this data broken down by metro. Really quite stunning work.
I also calculated some ratios and fractions, like change in GDP 2004-2006 and 2004-2008, City/Metro population ratio, and immigration as a fraction of metro population change. I figured these might have something to do with the buzz in the economy of each metro, and would predict more traffic congestion, commute times, and transit use.
It turns out that the only correlations with a r-squared over 50% were “Commute Time-Transit Share” (we knew that already), “Transit Share-GDP 2006”, “Commute Time-GDP 2006” “Commute Time-Metro Population 2006”. Those graphs and their correlations are shown below. By the way, New York, the “Big Apple” is shown in Red, while Atlanta, capitol of “the Peach State” is shown in Orange.
The last, strongest correlation is actually the most interesting, if I am willing to ascribe causality to it. Large Metros have large commute times, because there’s more of each other we need to get through to get to our fershlugginer work. Or so the story might go.
So I’m still missing something. These are not strong correlations, and I’m soon going to lurch over into ANOVA if I’m not careful. There might be interesting differences between cities above and below those trend lines, for example. I’d like to look at the job markets, income, and especially road and transit networks for these towns. My next post on this topic will depend on my finding those data.
Monday, I’ll probably be writing about crime near transit stations.
Well, time for our blended commute.