Crime analysis and the census|
Chapter 5: Synthesis and Applications
Postings on the Crime Mapping Research Center's listserv indicate that the manipulation of census data is a confusing obstacle for many analysts. This is not surprising, since census data and census geography are somewhat complex, or can be, depending on what you are trying to do. Among the most problematic areas are linking census geographies to data and misunderstandings relating to what can reasonably be, and not be, done with
specific census "counts" and variables.
For example, what measure of income should be used? The census contains data on family income and household income. Families are subsets of households, so these data for families exclude single
persons living alone or unrelated persons living together (who constitute households). Family income will be higher than household income because the average family has more wage earners than
the average household. Normally, because it is more inclusive, the preferred measure is household income.
A second question arises: Is it best to use median income or mean income? The mean has the advantage of being able to be manipulated mathematically, but in a skewed distribution, such as is found with income, the mean may be "pulled" to
the right by the very high values of the extremely rich (the so-called Bill Gates Effect!). For such skewed distributions, the median is usually preferred since it gives a more accurate description.
Detailed treatment of these questions is beyond the scope of this guide. An excellent review of issues in census geography and the analysis of change using census data can be found in Myers (1992).