[Contributors: Erik Feiereisen, Li Xie, Grace Giffin]
Because we are in Boulder, and because the wonderful librarians at the CU Boulder Map Library were best able to help us with historical Boulder maps, we decided to do our first mapping experiments on the city of Boulder (even though our first overall data explorations were on Weld County.)
The census data we are using is publicly and freely available, but it doesn’t have identifying information – so no names or addresses. Still, we really wanted to put some of these people on a map. We were curious about whether different ethnicities lived in different neighborhoods, whether we could see any occupational patterns, and so on. And it’s just cool to make a map. So we were in for some manual transcription work (in which we got a ton of help from the students in Kathleen King‘s history class at the Skyline High School, by the way – thanks!!)
The original census sheets of course contain both the names and the addresses we wanted. And fortunately, they have been digitized – at least we didn’t need to dig into microfilm!
Below is an example of the census sheet that we transcribed our data from.
It was very hard to read the freehand, so luckily there were transcriptions that showed up when you hovered your mouse over each box.
This census sheet is in the HeritageQuest genealogical database, which is what we used – lots of public libraries have access to it, and so does the Boulder Public Library, which is how we accessed it. We selected two census districts to work on – one near the university and the other closer to downtown – and manually entered the address, age, relationship to head of household, marital status, race, sex, occupation, and the individual’s and their parents’ birthplaces for about 900 individuals. We proceeded household by household, but we only input people over 16 years old – we figured that young kids would likely not tell us that much, as they didn’t generally have an occupation, for example, and other household members were more likely to be born outside Boulder.
So then we ended up with a spreadsheet like this:
Once we had a good selection of people on the sheet (or rather, once we ran out of time – the selection is really in no way representative or a carefully selected sample, but rather just contains the first few census sheets of two Boulder districts), we tried our hand at geocoding (finding out the geographical location of) these.
The first passes we made got some addresses entirely wrong – unsurprisingly, some street names have changed since 1920! For example, the street that’s now Canyon used to be called Water Street. But actually, there were surprisingly few changes for the downtown area at least, and our allies the map librarians were able to help us figure those out.
For geocoding & mapping, we used two different services: Google Fusion Tables (which, it turns out, will be discontinued in another year), and BatchGeo. Google FusionTables allows one to geocode and map up to 1,000 addresses; BatchGeo lets one do only 250 per map under its free plan.
The next pages in this section give a little bit more detail on what we found with each.