Where Harvard people are getting census data (as of 08/2009):
Steve Melly of HSPS reports good luck downloading Summary Files directly from the census ftp site: http://www2.census.gov/ and using the access templates to convert these to tables.
Sumeeta Srinivasan has had good luck getting data tables at the tract blockgroup and block level from American FactFinder. And hearing this, I did get it to work. Using “Custom Map -> Geo within Geo” you can dump put census data table by table. I find that it hangs a lot. One drawback of this is that you must download one table at a time (A table being more or less a report on the answers to a single question on the census form.) This is fine – especially if you know what you want, but can be a headache if your mission is more exploratory.
Guoping Huang points out that Factfinder will be overhauled for 2010, and so it may get better.
A few people are using geolytics. Pam Waterman and Jamie Hart of will actually extract data for people in their research cluster (I didn’t have the heart to ask if they would mind adding 500 GSD students to this cohort!) Scott Walker at the Map collection also uses it and finds it frustrating as I do. At the GSD, our library has bought a network site license for most of the geolytics products and I have mounted them on a shared volume and tried to make it so students can use the from their computers see www.gsd.harvard.edu/gis/manual/censuscd for details. But, as you can see by my docs, this is not the most straightforward thing in the world. Each of their products is a bit different, and they do not run on windows 64 bit, and to make these available on the net requires some hackery. In spite of this, I think I am going to continue to support this for use in our environment because geolytics is the easiest way to browse through all of the tables in a particular census year and pick and choose the data elements that you want into a single table.
Steve Melly points out that while Geolytics is somewhat easier than the official census sources, systematic errors in have been discovered in at least one geolytics field, and so for serious work, one should seriously consider doing the extra work to go with primary source material.
Jeff Blossom and Giovanni Zambotti at the Center For Geographic Analysis recommend the ArcGIS business analyst package which can be used in the CGA lab. This has advantages in that there are lots of non-census demographics (for example lots of market analysis stuff from Claritas) that can be easily obtained from there in addition to the most useful census stuff. I need to go check this out.
Guoping Huang also points out that the ESRI Maps and Data CDs have some of the most commonly requested census data. Census geometry can be downloaded from ESRI’s free data portal, and also from our own Harvard Geospatial Library.
Wendy Guan reports that CGA will be developing a FAQ about census data on their web site soon.
Last year Harvard Libraries purchased a license for Social Explorerhttp://www.socialexplorer.com.ezp-prod1.hul.harvard.edu allows downloads of all sorts of data down to the blockgroup level, including a lot of historic census data and estimates that extrapolate from the last decennial census. This looks a lot like American Factfinder, actually, except that it seems not to hang as frequently. Too bad it does not go down to the block level.
So I guess the answer as to the best way to get census data is “It depends.” I think that my answer for GSD students may be to use the ESRI Maps and Data CDs and if they need more detail in the attributes they should try social explorer or American factfinder. I still think that geolytics is worth digging into if you want to pick and choose lots of data from lots of census tables just to do some exploratory data analysis – but perhaps it is best to go back to the original source (e.g. ftp.census or American FactFinder once you have figured out what it is that you need.
As for strategies for moving forward into 2010… I think that I may stick with the strategy as discussed in the paragraph above, and also add a few more Geolytics products to our site license. I presume that all of the other options will be updated and will be no worse than they are now. Perhaps the Bureau of Census may even surprise us. I tend to agree with Scott from the Map collection that it may be worth watching the progress of subscription based web products like the geolytics and simplymap to see if any of these will be better than Social Explorer going forward – and if so we might join forces and purchase a Harvard wide site license.
- This information was compiled by Paul Cote, GSD
no links
no file attachments