I’ve been fielding calls from local reporters about this study…which I’ve quickly read…but I’m unclear about their SOI database. They only cite economists on the general concern with how “place” shapes intergenerational mobility (!!) but say that they compare “where” the child was located prior to age 13 and after 25 with parents income. All these incomes are essentially ranks of parental and child incomes using a rank correlation as the measure of mobility. Svalagostoga propagated the term “permeability” to reflect such measures but there are of course many others since his time.
I recently posted something on my other blog about Essential Python Geospatial Libraries, and thought that it might also be useful to the Spatial Demography community. The following is a list of Python packages that I (and others) regularly use for handling, exploring, analyzing, and generally mucking about with spatial data.
pandas – For data handling and munging. This is an extremely powerful tool for working with data in a spreadsheet-like format. If you’re familiar with R data.frames, then you’ll love pandas.
shapely – For geometry handling. This is the de facto package for geometry handling and manipulation.
cartopy – For plotting spatial data. This is a pretty new package, and is actively being developed. It makes for really nice publication quality maps already, and plays really nicely with shapely geometries and matplotlib (see below).
rtree – For efficiently querying spatial data. This is a relatively simple spatial index package that really speeds up spatial bounding box queries.
nodebox-opengl – For playing around with animations. Everyone loves animations, and this package makes it pretty easy to create some simple, interactive animations.
statsmodels – For models and stats in Python. The group behind this project is trying to make Python just as easy and powerful for stats as R currently is… and they are making excellent headway already.
numpy – For pretty much anything that involves arrays. The is probably the most important package for data analysis in Python.
geopy – For geolocating addresses and things like that. It is a small wrapper around various web-based geocoding APIs.
ipython – For a wonderful interactive Python environment in which to play. It makes working with Python a true joy.
ogr/gdal – For reading, writing, and transforming geospatial data formats. This has all sorts of nice ways to work with geospatial data, though fiona (see below) is much nicer for reading and writing different geospatial formats.
pyqgis – For anything and everything GIS. This is the Python API for Quantum GIS. You can work directly within QGIS via its intergrated Python console, or create standalone GIS apps using this powerful Python package.
fiona – For making it easy to read/write geospatial data formats. Really, really nice API for reading and writing GIS formats.
matplotlib – For all your plotting needs. The de facto plotting library for Python. It does everything from scatterplots and histograms to choropleth maps to complex interactive visualizations.
networkx – For working with networks. Very nice for visualizing (integrates with matplotlib) and working with network data.
pysal – For all your spatial econometrics needs (and more). This is a really great package that is also under constant development. It covers everything from exploratory spatial data analysis (ESDA) right up to heavy duty spatial econometric models.
descartes – For plotting geometries in matplotlib. This is a nice, clean, GeoJSON style data helper for matplotlib and shapely.
geographiclib – For solving geodesic problems. It also converts between geographic, UTM, UPS, MGRS, geocentric, and local cartesian coordinates, and even does geomagnetic field calculations.
pyshp – For reading and writing shapefiles in pure Python. Nice API and works great.
pyproj – For conversions between projections. This package provides an intuitive way of interacting with the Proj4 library for transforming data between coordinate reference systems.
It is also probably a good idea to mention scipy, which is a Python-based ecosystem of open-source software, and includes things like numpy, matplotlib, ipython, pandas, and much more.
The best way to start exploring and working with these various packages is to create a Python virtual environment, and use pip or easy_install to grab the latest versions.
Earlier I had posted on the issues raised in the Santa Barbara (UCSB) meeting of spatial demography specialists regarding ecological vs. grounded approaches to the social community concept. The current issue of Rural Sociology (78, 2, June) has a nice review of much literature from sociology by Peter Fairbrother et al. in “Creating ‘Community’: Preparing for Brushfire in Rural Victoria”. It’s worth reading just for this summary, IMHO.
Poverty Segregation in Nonmetro Counties: A Spatial Exploration of Segregation Patterns in the US
By: Johnelle Sparks, Corey S. Sparks and Joseph J.A. Campbell
Abstract: Most research on segregation focuses on racial residential segregation in metropolitan statistical areas and typically uses a-spatial measures of segregation. What is less clear is if segregation measures operate in a similar fashion in nonmetropolitan areas and if spatial patterns exist for poverty segregation in nonmetro counties. The purpose of this research was to examine multiple dimensions of poverty segregation in the United States the period 2006-2010 for metropolitan and nonmetropolitan counties. Data for this analysis come from the 2006-2010 American Community Survey 5 year estimates, the 2000 U.S. Census of Population and Housing, Summary File 3 and the USDA Economic Research Service. Four different measures of poverty segregation were calculated, including both aspatial and spatial measures. A nonparametric Kruskal-Wallis test was used to test for variation in the segregation indices across metro and nonmetro areas and spatially autoregressive models were used to examine the socioeconomic correlates of poverty segregation. Results indicate significant variation in poverty segregation patterns in metro and nonmetro counties in the US, and nonmetro counties outside of the South have significantly lower levels of poverty segregation. This research adds to the literature by exploring patterns of metro and nonmetro poverty segregation and measuring different dimensions of segregation with an explicit spatial referent across counties in the contiguous United States in an effort to note differences in how segregation works across rural and urban places.
Stephen and I have a paper, entitled “Spatial Polygamy and Contextual Exposures (SPACEs): Promoting Activity Space Approaches in Research on Place and Health,” in American Behavioral Scientist. The abstract and full text is available here.
Briefly, this paper mainly argues that current research approaches to individual exposures to environment are confined to arbitrarily defined boundaries and overlook the fact that individuals’ daily life crosses these boundaries (the concept of spatial polygamy). To address this issue, we suggest that future social and health studies should take human mobility into account and develop innovative methods to capture the complexity of individual exposures to environment. Several recommendations are provided at the end of this paper.
We welcome any feedback or comments on this paper and hope to spark discussions that will advance spatial demography.
I’ve been working for several years on a project that involves geocoding all the residents of several cities in 1880. Here is a historical puzzle. In the attached map of a portion of Baltimore, individual buildings have been coded as predominantly white or black. Almost always “predominantly” means all the residents are the same race.
I am used to thinking of neighborhoods as some sort of polygon, extending from a central core along the streets in every direction. In this case we see evidence of linear neighborhoods, where the “neighborhood” is defined solely by racial composition. Three north-south streets are nearly all black for many blocks (Dallas, Bethel, Durham),. The parallel streets are nearly all white.
First question: where have you seen this linear pattern before? I have not.
Second question: what is the source of this pattern on these particular streets? There must be some specific history to it.
I’ve made the red dots smaller on this map so that the black resiential pattern stands out more clearly. This section of the city was majority white.
The Final Report from the Future Directions in Spatial Demography may be of interest to the readers of the new journal, Spatial Demography.
The Future Directions meeting was held in December 2011 in Santa Barbara and included participation from 40+ scholars from demography-related disciplines including geography, economics, sociology, anthropology, political science, and rural sociology. The Final Report was released in April 2012 and made available at the following website:
The purpose of this forum is to highlight the tools of the trade, our methodological toolbox, if you will. With so many scientists in so many disciplines contributing to the area known as “Spatial Demography”, we all have our old stand by routines, our tricks and our tips for new researchers. This is how I see this column evolving.
I will try to routinely post how-to guides to various techniques of spatial analysis and spatial statistics using the tools I know. This should include primarily open source computing applications, with lots of annotated code, but occasionally commercial or proprietary software will be highlighted as well. The first column for my area, for example, shows how to use the free R software package to read ACS data downloaded from American Factfinder, join it to a shapefile and conduct exploratory spatial data analysis using some standard methods. Future columns will continue to use R for various other programming and analysis examples and expand into other languages and platforms as time goes by.
While I’ve managed to come into contact with lots of different platforms and software over the past decade, I welcome those among us who make a habit of writing code, developing software, or even like to tinker with every program under the sun that will open a shapefile to contribute to this area. I welcome brief discussion for the forum, but would also welcome longer pieces of 1,000-3,000 words for publication in the regular column for this area.
I am very excited to be involved in this journal dedicated to our particular area of expertise and interest and personally look forward to growing our shared knowledge base over the next few years.