[R-sig-Geo] US Census 2000 and 2010

[Virgilio Gómez-Rubio]

Dear Chris,

> that includes all the data, but what I would really like are some pointers
> to scripts that will get just the data I need straight from the Census and
> read it in so I can use it. I am interested in both the SF3 variables and
> the Tiger/Line files.

In addition to all the documentation in the USCensus2000 packages, you
may find this talk by Zack Alquemist at useR!2010 interesting:

http://www.r-project.org/conferences/useR-2010/slides/Almquist.pdf


The whole suite of USCensus packages is discussed therein and he points
to some future directions. I'd say that these packages could be a
starting point to develop code to get the data from the 2011 Census...

> 2)Has anyone worked with aligning census geographies at a fine scale (e.g.
> matching 2000 Census blocks to 2010 block groups). Obviously there is a lot
> of change at a fine scale and I was wondering if anyone had worked with the
> Census Crosswalks to try and bash out a workable set of spatial units for
> cross census comparison?

I do not think that this a simple issue with a simple answer... If you
have misaligned regions you will have to make a decision about how you
aggregate/split. There is a discussion about this in the book by
Banerjee et al. (on hierarchical modelling and analysis for spatial
data).

> These are both big jobs and will likely be the focus of my work for some
> time, but if anyone has some pointers at the outset, I would love to not
> reinvent the wheel.

You may want to look at the literature on small area estimation. Look
for the paper Muggly et al (2000, "Fully Model-Based Approaches for
Spatially Misalign") and then you can look for posterior papers that
cite that paper.

Hope this helps,

Virgilio