I have been using the census block shapefiles and tables for 1990, 2000, and 2010 available from NHGIS without really understanding what was going on, because NHGIS makes it so easy. NHGIS even offers a choice of shapefiles for 2000 because Census provided to versions based on different methods. I’ve assumed that for whatever reason, the blocks for 1980 and before were not digitized and hence NHGIS cannot provide it.
I was not clear on why blocks for the territories do not show up on the list. I assumed that either Census did not create all the data, budget limitations forced omission, or a Caribbean initiative elsewhere did it.
Now I am trying to get the same data for Puerto Rico, and possibly other territories. So I probably need to finally understand what is really going on. Perhaps I need to duplicate for PR what NHGIS did for the states, or perhaps it is impossible. Or maybe someone else did it. I am hoping you can fill in the gaps in sources and my understanding.
I downloaded without issue the TIGER shapefiles for 2000 and 2010 that Census offers at TIGER/Line Shapefiles . Census claims that I will be able to find the tables to associate at data-dot-census-dot-gov I expect that joining those tables will be a bit more tine-consuming as it lacks your GISJOIN field; but not too bad once I find the tables comparable to the csv files NHGIS provides.
But what really stumped me concerns the 1990 blocks and—while less important—the blocks for 2000 using the 1990 approach. There is a statement that Census 2000 is available in both formats, which looks promising. But when I downloaded the files from https-www2-dot-census-dot gov/geo/tiger/tiger2k/, the format is not anything that ArcGIS10 or ArcGIS pro recognize. For 1992 Census warns
The 1992 TIGER/Line Files serve as a link between 1980 and 1990 geography. These files are in an early TIGER/Line format that generally does not work with shapefile conversion software….Technical support is not available for these files.
My guess is that 1992 is the vintage of the digital product for 1990 blocks. A cursory web search suggests that others have been stumped by this obsolete format. Should I infer that NHGIS managed to read those same files and convert them to shapefiles? The discussion of similar issues at GIS Files | IPUMS NHGIS does not delve into this in detail, so I am wondering whether the difficulty of doing so is one reason why NHGIS did not do so for Puerto Rico. If this is really time-consuming, I can just drop the matter as impractical, but I wanted to check with people who would know.
In addition, can you shed light on the statement by Census that the 1992 provides a link to 1980 geography? Does this mean that there is a digital product for the 1980 blocks as well which, to date, simply has not been converted to a format that most people can use?