National censuses, such as the decennial census of the U.S., are among the richest sources of attribute data. Attribute data are heterogeneous. They range in character from qualitative to quantitative; from unranked categories to values that can be positioned along a continuous number line. Social scientists have developed a variety of different measurement scales to accommodate the variety of phenomena measured in censuses and other social surveys. The level of measurement used to define a particular data set influences analysts' choices about which analytical and cartographic procedures should be used to transform the data into geographic information.
Thematic maps help transform attribute data by revealing patterns obscured in lists of numbers. Different types of thematic maps are used to represent different types of data. Count data, for instance, are conventionally portrayed with symbols that are distinct from the statistical areas they represent, because counts are independent of the sizes of those areas. Rates and densities, on the other hand, are often portrayed as choropleth maps, in which the statistical areas themselves serve as symbols whose color lightness vary with the attribute data they represent. Attribute data shown on choropleth maps are usually classified. Classification schemes that facilitate comparison of map series, such as the quantiles and equal intervals schemes demonstrated in this chapter, are most common.
The U.S. Census Bureau's mandate requires it to produce and maintain spatial data as well as attribute data. In Chapter 4, we will study the characteristics of those data, which are part of a nationwide geospatial database called "TIGER."