Whenever I see new data on cities, I'm always tempted to match them to the cities' weighted densities, if for no other reason than no one else does it. And so with the Texas Transportation Institute's latest report on city congestion. TTI found a wide range in hours lost due to congestion per year -- e.g., in 2007, Lost Angeles drivers lost an average of 70 hours per year to congestion; Cleveland drivers, just 12. (The relevant TTI table is here (pdf).) Does weighted density partly explain this variation?
The answer is "No," based on my admittedly simplistic analysis.
Below the jump I have three charts plotting, for 33 cities, weighted density, standard density and total population against hours lost per traveler to congestion. (The 33 cities include the 31 largest urbanized areas -- excluding New York City, which is always an outlier -- and Austin and Honolulu.)
As the scatter plot shows, weighted density explains virtually none of the variation in congestion (adjusted R2 = .06). Standard density explains a bit more (adjusted R2 = .19). And total population, a bit more than standard density (adjusted R2 = .28).
I expected weighted density to be a key explanatory variable. And it would have been more exciting to announce that weighted density matters, either one way or the other. But the first chart suggests it does not -- a city with a high weighted density is no more likely to be highly congested than a city with a low weighted density (and vice versa). While not particularly exciting, the fact that weighted density doesn't seem to matter is important, too.
As I routinely say when I put out stuff like this, don't take it too seriously. This is a small sample. More importantly, some very smart people have employed some very sophisticated methods to analyze the variation in congestion. The absolute size of the metropolitan area population is a key factor. The rate of growth matters a lot, too, since fast-growing cities have trouble keeping up with rising demand for roads. Higher standard densities are associated with increased congestion, although the strength of the association is disputed, I think. And, of course, economic vitality plays a big role -- peak-period congestion means a lot of people with somewhere to go, which implies a healthy economy. (Anthony Downs has a good discussion of these issues in Chapter Three of his book, Still Stuck in Traffic, most of which Google Books has helpfully published on-line.)
(I've uploaded to Google docs a spreadsheet showing total population, weighted density, standard density and annual hours per traveler lost to congestion.)
