Questions for the Census Bureau?



Next week I will be traveling to Washington to meet with someone at the US Census Bureau. While I’ve had several phone calls with people at Census, this will be my first face-to-face meeting with someone who works there.

As you might imagine, I have a long list of things I want to cover! In fact, I’m sure that we won’t be able to cover everything in a single meeting.

One thing I’d like to make time for is questions from the larger R community. At this point several thousand people have taken my free course Learn to Map Census Data in R. And many users of my open source projects also work with Census data.

So if you have any questions about data from the US Census Bureau, or have any documentation or feature requests for their data, feel free to leave them as a comment here. I’ll do my best to pass them along.

skjainmiah says November 14, 2016

All the best sir, great chance to prove what you are at high end. I hope some or other time you would also visit our Rayalaseema University . Thank you sir.

timkiely says November 14, 2016

Ari – In general, I find it very difficult to simply find the tables I’m interested in from ACS. For example, if I wanted to look through the different variables they have for “rent” and I issue a term search using the API, I get thousands of results back which I have to carefully read through one by one, and the table names are long and complicated. Then there is the issue of availability by different spans and ending years. Overall, it would be great if the data were simply more searchable.

Alex Thompson says November 14, 2016


I think basic, super simple, clear statistics in an Excel file or, better yet, csv should be available in an incredibly painless way. Namely, I understand that American Fact Finder exists, but sometimes it needs to be simpler than that. For example, if I Google “census data by county” this absolutely should not be the thing that comes up :

Let me know what you think.


andrewdavidmitchell says November 15, 2016


I’d appreciate learning more about how their data is collected and verted in the first place.

Under what circumstances is a statistical model implemented in place of a true census?

Additionally, would they consider taking a more active role in publicizing their findings to the general public?

Harris says November 15, 2016

This isn’t strictly related to the Census Bureau, but I am wondering if you know of free, open-source software that would allow me to generate shapefiles to use in R.

Nunya Dempleter says November 15, 2016

I’d like to know the most direct way to import a subset of census tract shapefiles into R for the purpose of making choropleths illustrating demographic variables.

Jeff says November 16, 2016

Less of a question a more of a request and observation. I noticed that the page has many visualizations but no data tables. While there are many interesting visualizations, and they may prompt some creative spin-offs, the main focus should be on the data and code to generate visualizations, not just the final image. There seems to be more focus on “tools” and less focus on data availability. Perhaps a Github repository would be a great place to compile code used for the visualizations.

Joel says November 16, 2016

One of the things that has become increasingly frustrating with the move to ACS-based core statistics (for me at least) has been trying to figure out which data elements are available without suppression at the lowest levels of geography, in particular block groups in the 5-year ACS summarizations. As readers of this blog will likely appreciate, this is particularly an issue with thematic mapping and spatial modeling applications, where in my experience there are often substantial voids in whatever data element you are looking for, at least in some of the more specialized topical areas (commuting, occupations etc). A related issue is that the technical documentation on which tables are available at which levels of geography would nominally show these layers as ‘available’, at least at an overall table level (ie., this layer exists for SOME geographies at this level).

With the understanding that these suppression rules are deterministic by nature, and there is not much that can be done about them by nature:

Is there some sort of meta-documentation on invocation of the suppression rules on a table (or table and data element) basis, this would at least allow the analyst to see which layers are actually complete from a basic data-resolution standpoint. Ideally (for me, at least) this would be bucketed by Metro area, so that we could see in a specific use context whether or not the data is actually available, and how many geo-units are suppressed …this would be very useful…haven’t found it if this type of document/file exists…

Amelia McNamara (@AmeliaMN) says November 16, 2016

Like timothyjkiely, Alex Thompson, and Joel, I would like easier access to the tables I am looking for, some basic things that people (especially students) request commonly, and a better way to know if they will be available for the geography I am interested in. My rant on Census data has some more thoughts: Is your meeting related to the rollout of ?

    Ari Lamstein says November 16, 2016

    Thank you for posting this! The meeting is unrelated to the link you provide, and is very informal. I’ve been corresponding with a few people there via email and phone for a while now. And since I’ll be on the east coast for Thanksgiving, I just wanted to drop by and put a face to a name. I was delighted when they agreed to make time for me.

M says November 28, 2016

Census already has a page that provides SAS users (& Excel users) helpful resources related to the American Community Survey, like data formatted for SAS and sample code. Census should provide similar resources for R users.

Comments are closed