Chapter 11 Review Questions

§11.1 - §11.3 Ecological Cross-Sectional Studies

  1. Is "crime rate of the neighborhood in which you live" a person-level or aggregate-level measurement?
  2. Farr derived a statistical model that was nearly perfect in predicting cholera rates. He used this information to support the miasma theory of cholera transmission. Why was Farr so far off base?
  3. Optional: Match the term with its description. Terms: integral variable, contextual variable, contagion variable Descriptions: (1) An aggregate-level variable that derives from compilation of individual attributes.(2) An aggregate-level variable that affects virtually all members of a group.(3) An aggregate-level outcome variable that affects future occurrences.
  4. Figure 11.5 (p. 198) revealed a positive association between hospital size and accidents rates. The author attributed this to the lower quality social interactions at large hospitals. Suggest an alternative explanation.
  5. True or False? Whether a measurement is longitudinal or cross-sectional depends on the date of occurrence in relation to data collection.
  6. What are notiones vulgares? 
  7. Table 11.5 (page 201) reveals a positive association between psychosis and low SES and a negative association between neurosis and low SES. Explain how biases might explain  these findings.
  8. Match the bias with its description. Biases: detection bias, reverse-causality bias, prevalence-incidence bias Descriptions:  (1) long-duration cases are weighted more heavily than short-duration cases  (2) the exposure makes it more likely for cases to be diagnosed but does not cause the ailment    (3) the disease causes the exposure (not vice-versa)

§11.4 Cohort Studies

  1. Explain why mortality rates in open populations are cross-sectional and not longitudinal. 
  2. Explain why Frost's generational cohort studies is the link between infectious and chronic disease epidemiology. 
  3. Where is Framingham and why do we care? 
  4. Who are Doll & Hill?
  5. Describe the difference between a prospective cohort study and retrospective cohort study. 

§11.5 Case-Control Studies

  1. Suppose a disease occurs at the rate of 1 per 1000 person-years. How long would it take to accrue 100 cases in a population of 10,000 people?
  2. Optional: Levin's  data (p. 209) has an odds ratio of 2.5 for smoking and lung cancer. Other studies have estimated the odds ratios to be at least 10. Propose an explanations for Levin's underestimate.
  3. This is another name for the cross-product ratio in a 2-by-2 table.
  4. Controls is case-control samples provide an estimate of this. 
  5. Select the best answer: The odds ratio from a case-control study is stochastically equivalent to a(n): [(a) incidence (b) prevalence (c) rate ratio (d) rate difference].
  6. Optional: When there is an equal number of case-positive and case-negative discordant pairs in a matched case-control study the odds ratio is equal to ________. 
  7. Optional: T/F: The information from concordant-pairs in matched case-control studies is ignored when determining the odds ratio. 

Last update: 11/30/2008  

Key