Originally Posted by
spare_wheel
ACS counts are based on unbiased surveys and modelling of the decennial census. The census is run by highly a-political math wonks. Traffic departments/bureaus, on the other hand, are highly politicized and select count sites not based on a statistical model but based on political and/or arbitrary criteria.
Since you agree that ACS numbers should reflect a trend in cycling then what exactly is your explanation for discrepancy in numbers between the ACS and the BBC?
I'll let the DOT and the Bureau of Transportation Statistics explain that discrepancy. they do a far better job than I would.
Here's some of what those "math wonks" say about their own census data as it pertains to bicycle pedestrian counts.
Originally Posted by US DOT and Bureau of Transportation Statistics
Some commonly recognized limitations of census data for pedestrian and bicy- cle analysis are listed here:
● Only work trips are included. Thus, if census data are used to represent overall levels of walking and bicycling, these must be assumed to be in some proportion to work-trip use. Work trips make up less than one- quarter of all trips, so nonwork trip-making patterns are not captured by the census.
● The data may not represent “normal” pedestrian and bicycle work-trip mode use. The census is conducted the first week in April, and respon- dents are asked to report their most frequently used mode in the past week. Mode use may fluctuate depending on weather conditions, and occasional bicyclists or walkers would not be included. Also, trips by mul- tiple modes (e.g., a walk to the subway) cannot be determined.
● Data are not available at both the disaggregate level and a high level of geographic detail. PUMS disaggregated data are only available for small samples at the level of the city or county, and the application of PUMS data to pedestrian and bicycle analysis has been very limited...
For more info go to this site:
DOT/BTS document for more information on the gaps and limitations of census data and how the "math wonks" use an amalgam of bicycle counts, ACS and other census data to determine needs, usage and for design and effectiveness of various infrastructures. The "math wonks" do not dismiss one set of data, bike counts, census figures etc but instead weigh
all the data with the pragmatic realization that the collection of data may have inherent flaws but can be cross checked to determine the most accurate measurements, which is all I am suggesting.
I will ask you one more time and then I will give up but please show me in the Boston Bicycle Count which areas have infrastructure and/or which do not. You stated with certainty that the count was skewed in favor of infrastructure. That the counts were done only at places with new infrastructure. I'm really curious as to how you support that contention.