Tuesday, January 22, 2013

GIS, reflections on data collection, and the Brazilian census

I recently took my first introductory GIS course, where the final project aimed to tie together all of the spatial analysis tools we had surveyed, but had yet to sufficiently chew on. I'll likely expand on this post later to extol the benefits of a course in GIS for researchers and planners (contingent on the pedagogy and experience of the instructor, clearly), its use in training capable decoders and critics of the visual information with which we are inundated daily as modern men and women, etc. But for now I'd like to touch upon the more immediate, practical issue that all novices, like myself, will confront in the initial stages of any research design: bad data, old data, or an apparent lack of any data.

Let's not be sloppy, simply because we're foreign and not "in the know." 


This post was actually inspired by the intraweb trails of a French doctoral student. He too had been posting questions to various Brazilian forums, inquiring in  Portuguese about where to find shapefiles of the Metropolitan Region of S
ão Paulo for his dissertation. Several geographers conventionally pointed him toward Centro de Estudos da Metropole, a site which provides very useful information and geographic data, and is sponsored by various institutional arms of the State of São Paulo. This site is highly recommended if you are researching any issue related to cities and urban areas in Brazil. 

However, the shapefiles supplied by CEM are mostly stale shelf stock from the 2000 census. Census tracts have been redefined or have been newly created in many parts of Brazil since then. In addition, many of these shapefiles are already joined to tables with basic census demographics, created for practitioners and planners learning TerraView, the Brazilian/Latin American opensource mapping platform created as an alternative to ArcGIS (which is often too costly and too "elaborate" for the purposes and, arguably, the skill sets available to many Brazilian municipalities). Hence, these outdated shapefiles may also prove unwieldy and loaded with fields that you might otherwise not need, depending on your research question. 


Well, unfortunately, it appears that our anonymous cyber colleague settled on the 2000 census data, since there is no evidence in the forum that he was redirected to the main source for the 2010 census: The Brazilian Institute of Geography and Statistics (IBGE) website. His digital footprint leads us to assume that he handed in a doctoral dissertation in 2011 to make a contemporary observation about Brazil, using really old data from 2000.

Perhaps this was an isolated case? Perhaps foreign researchers are rarely encumbered by language barriers, research cultures, or their own natural unwillingness to undertake the frequently arduous task to find up-to-date and, importantly, clean data? My own skepticism remains in tact...


Oh yeah, and I read a study authored by an unnamed scholar who also works for the World Bank, who had also employed the 2000 census to make a non-comparative, contemporary statement about Brazil...So it's not just a junior researcher "thang." And probably not just a Brazilianist "thang," either. 

Where and how to find shapefiles and demographic data for the most recent Brazilian Census (2010) (in English)

All census data (statistics, shapefiles) can be downloaded directly from The Brazilian Institute of Geography and Statistics (IBGE), following a presidential decree signed by Lula to make such information available. The following example is of São Paulo, but other cities could be found by querying a different name. 

The Brazilian Institute of Geography and Statistics
http://www.ibge.gov.br/home/default.php

Shapefiles of census tract boundaries:
The Brazilian Institute of Geography and Statistics - Geociencias

1. From the IBGE homepage, navigate to “download”.
2. On the left-hand panel, click on “geociencias”.
(http://www.ibge.gov.br/home/download/geociencias.shtm)
3. Follow the path “malhas_digitais/censo_2010/”.
4. To download shapefiles of 2010 census tracts of the São Paulo Metropolitan Region, click
“setores_censitarios” and download “sp.zip (14942 kB)”.
5. To download shapefiles of favelas of Brazil, click “aglomerados_subnormais/shape” and download
“aglomerados_subnormais2010_setores.zip (6555 kB)”.

The following shapefiles for metropolitan region of São Paulo and favleas:
• 35DSE250GC_SIR.shp
• 35MUE250GC_SIR.shp
• AglomeradosSubnormais2010_SetoresCensitarios.shp

Excel spreadsheets containing 2010 census data of the São Paulo Metropolitan Region:
The Brazilian Institute of Geography and Statistics - Estatistica

1. From the IBGE homepage, navigate to “download”.
2. On the left-hand panel, click on “estatistica”.
3. Follow the path “Censos/ Censo_Demografico_2010/ Resultados_do_Universo/
4. Download “Base_informacoes_setores_2010_universo_SP_Capital.zip (165470 kB)”.

The following data sheets contain more commonly used information, such as household income, age, literacy. The guide on how to read the census spreadsheet is also located in the 2010 census download: 

• DomicilioRenda_SP1
• Domicilio01_SP1
• pessoa01_sp1
• Basico_SP1
• pessoa13_sp1