Question 1078989
I might consider a cluster sample, using for the cluster all the businesses and schools in the city and then taking a random sample of the business, then getting the response from every person in that business.  This would have a strong relationship to the use of a commuter rail station.