However, the analysis is easy to do. The instruction I inserted was:
if(tc == "BigCity") i01 = i01[tv$ipop>2000 &tv$urban == "C"];
I also did MedCity, with pop >500,000. Joseph's analysis is here
Trends
| Trend | BigCity | MedCity |
| Number | 205 | 562 |
| 1901-2005 | 0.1154 | 0.08848 |
| Trend se | 0.007321 | 0.007264 |
| 1979-2005 | 0.3073 | 0.291 |
| Trend se | 0.05195 | 0.05094 |
Plots
 | 
| |
 | 
|
Posts
The trend of the last 30 years doesn't look too different, that's for sure. Here's the graph I produced of 12-year running averages, comparing cities greater than 2 million, and towns with population less than 15,000.
ReplyDeleteTo see the most pronounced effect possible, I'd suggest cities bigger than 10 million - in the raw data, of course.
There should be 35 such stations.