Scenario: Childhood leukemia is positively associated with socioeconomic status (SES), so that wealthier families have elevated childhood leukemia rates (the mechanism of this association is unclear). It also is hypothesized to be associated with exposure to compound X. A researcher observes a map of county-based mortality rates for childhood leukemia from 1990 through 1995. The State has 102 counties, 22 of these contain injection wells contaminated by compound X.

Questions:

  1. In general do you expect socioeconomic status (SES) to be similar within neighborhoods? Between neighborhoods? At the county level?
  2. The researcher decides to test for a difference in means between mortality rates in the 22 vs. the remaining 80 counties. The proposed test assumes independent observations. What frequently found characteristic of spatial data violates this assumption?
  3. The injection wells tend to be located in counties of lower SES. How would this impact the results if the hypothesized positive relationship between compound X and childhood leukemia is true?
Return to Scenerios