The article says they do the survey on a particular day each year whereas the data set is less specific on that fact. Surely they do the survey over multiple days...? If available, there must be a ton of citibike usage data one could use to verify the citibike stats collected by this survey and generally validate this data a bit.