• The whole data for this project was gathered from the 2000 U.S. Census (American Fact Finder). It has advantages and disadvantages to rely on only one source. Without additional information from other sources you cannot try out whether there are any contradictions within the data but on the other hand you can be rather sure about the metadata, meaning that the data was probably processed under the same standards and circumstances. In this case it is an advantage that all my data is from the U.S. Census because it is an official source that is interested for itself in accurate and precise data. Therefore it is - relatively (errors may occur everywhere) - safe to rely on their data.

• The web site www.census.gov offers a variety of information for every level of accuracy (e.g. state, census tract, or block). Some of the data were collected from every single person whereas other are only from a 1-in-6 sample.
  

  
  
• It is easy to download the data you need, including the additional feature of transposing rows and columns which was really very useful in my case. But otherwise it was very time consuming to download data for 2,631 census tracts. It was impossible to do this at once. Instead I had to download small packages and put them together later. The process of downloading needed a whole working day!
  

  
  
• The shapefiles 'counties_merge.shp' and 'water_merge.shp' were downloaded from www.census.gov/geo/www/maps/ but could not be used in that state. Instead they had to be processed further, too.