## 2.4 Exercises

1. What is a general term used for methods that assess the association between an outcome and one or more predictors?
1. Correlation
2. T-test
3. Regression
4. Analysis of variance
1. True or false? A simple linear regression model has only one predictor.

2. True or false? The outcome in a binary logistic regression model can take on only two possible values (e.g., presence or absence of a disease).

3. True or false? Regression methods can estimate the extent of effect modification by controlling for confounding.

4. Which of the following are valid purposes for regression?

1. Controlling for confounding
2. Estimating the magnitude of effect modification
3. Testing an association
4. All of the above
1. Match the outcome type with the appropriate definition.
Outcome Type Definition
Nominal Has a small number of possible values.
Binary A categorical variable in which the levels have a natural ordering.
Continuous A categorical variable in which the levels do not have a natural ordering.
Event time Takes on non-negative integer values (0, 1, 2, …) and describes a number of items or events.
Ordinal Has a range of possible values and can take on any value in an interval.
Count Describes the time from a time origin to an event.
Categorical A categorical variable with exactly two levels.
1. Match the outcome type with the appropriate regression method.
Outcome Type Regression Method
Nominal Linear
Binary Cox
Continuous Poisson
Event time Multinomial logistic
Ordinal Binary logistic
Count Ordinal logistic

For the following research questions, the wording assumes the outcome is mentioned first. For example, “Is (outcome) associated with (predictor)”? For example, for the research question “Is the prevalence of Type 2 diabetes mellitus (T2DM) associated with median household income?” the outcome is prevalence of T2DM and the predictor is median income.

For each of the following research questions:

• What is the outcome type (continuous, binary, ordinal, nominal, count, or event time)?
• What is the appropriate regression method?
1. Is total cholesterol associated with weight?

2. Is time to ischemic heart disease associated with binge drinking?

3. Is seropositivity for COVID-19 (positive/negative) associated with occupation?

4. Is hospital length of stay in days following surgery associated with age?

5. Is body mass index category (underweight, normal, overweight, obese) associated with hours of vigorous activity per week?

### References

Alsuwaidi, Ahmed R., Farida I. Al Hosani, Shammah Al Memari, Hassib Narchi, Laila Abdel Wareth, Hazem Kamal, Mai Al Ketbi, et al. 2021. “Seroprevalence of COVID-19 Infection in the Emirate of Abu Dhabi, United Arab Emirates: A Population-Based Cross-Sectional Study.” International Journal of Epidemiology 50 (4): 1077–90. https://doi.org/10.1093/ije/dyab077.
Degerud, Eirik, Gudrun Høiseth, Jørg Mørland, Inger Ariansen, Sidsel Graff-Iversen, Eivind Ystrom, Luisa Zuccolo, Grethe S Tell, and Oyvind Næss. 2021. “Associations of Binge Drinking with the Risks of Ischemic Heart Disease and Stroke: A Study of Pooled Norwegian Health Surveys.” American Journal of Epidemiology 190 (8): 1592–1603. https://doi.org/10.1093/aje/kwab063.