EXERCISES

 

The first 2 exercises are from Hulley, p. 48.

Exercise 1:

Classify the following variables as dichotomous, nominal, ordinal, continuous, or ordered discrete. Could any of them be modified to increase power, and how?

  1. Sex.
  2. Age.
  3. Education (high school degree/no degree).
  4. Education (highest year of schooling).
  5. History of heart attack present/absent.
  6. Number of drinks containing alcohol per day.
  7. Depression (none, mild, moderate, severe).
  8. Percent occlusion of coronary arteries.
  9. Hypercholesterolemia (>250, <250).
Exercise 2:

The research question is, “Does body weight at age 1 year predict the number of drop-in clinic visits during the following year?” You plan a prospective cohort study, measuring body weight using an infant scale. You note the following problems while pretesting your measurements. Are these problems due to lack of accuracy, lack of precision, or both? Is the problem mainly due to observer, subject, or instrument variability, and what could be done about it?

  1. When babies are scared, they try to climb off of the scale and the observer must hold them on the scale to complete the measurement.
  2. If the baby is “squirmy,” the pointer on the scale swings up and down wildly.
  3. When you weigh a 10-pound reference weight on the infant scale 20 times, the mean weight is 10.01 ± 1.00 pounds (mean ± SD).
  4. When you weigh a 10-pound reference weight on the infant scale 20 times, the mean weight is 10.60 ± 0.01 pounds (mean ± SD).
  5. Some of the babies arrive for the examination immediately after being fed, whereas others are hungry; some of the babies had wet diapers.
Exercise 3:

Convert the following conceptual variables into operational variables.

  1. Compliance with medication.
  2. Good mobility of limb following hip surgery.
  3. High quality plain chest radiograph.
  4. Family dysfunction.
Exercise 4:

You have a questionnaire to measure anxiety in a study of mental health. This has been translated into Xhosa. To assess the quality of the questionnaire, it is tested against an in depth interview with the subject conducted by a Xhosa speaking psychiatrist (Table 1).

  1. Calculate some simple measures of the agreement between the questionnaire against the psychiatric assessment. What attribute of the questionnaire is being measured? How could you improve it?
    Table 1
        Psychiatrist assessment  
        High anxiety Low anxiety  
    Questionnaire High anxiety 20 35  
    Low anxiety 10 35  
            100
  2. Two trained interviewers administer the questionnaire to a number of the same subjects (Table 2). What attribute of the questionnaire is being measured? How could you improve it?
    Table 2
        Interviewer 1  
        High anxiety Low anxiety  
    Interviewer 1 High anxiety 40 15  
    Low anxiety 20 25  
            100
  3. What is the prevalence of high anxiety in this patient population?