What is effect modification?
When exposure-outcome association differs depending on a third variable
Give an example of effect modification.
High physical activity reduces type 2 diabetes risk by 70% in men but 30% in women — gender is an effect modifier.
What does an interaction represent statistically?
Evidence for effect modification in regression models.
What is a confounder?
third factor explains all or part of the association between an exposure and an outcome, obscuring the real effect.
What is a mediator?
- Variable that’s caused by exposure & in turn affects outcome
- Lies after the exposure in time
- Explains how or through what mechanism the effect occurs
Can a variable be both a confounder and an effect modifier?
Yes — it can distort the association and also change its strength across levels.
Can a variable be a confounder and a mediator?
No, Variable is either a confounder or a mediator, never both in the same analysis.
When should you test for interactions?
At final stage of multivariate analysis, focusing on a-priori hypotheses.
Why are formal tests for interaction underpowered?
Because large sample sizes are required to detect anything but strong interactions.
What indicates a significant interaction statistically?
A p-value < 0.05 for the interaction term.
What is the purpose of stratification?
To assess effect modification by running models within strata of the potential modifier.
How do you specify an interaction term in Stata?
- ## between two variables
- e.g., i.sex##i.smoke.
What does c. mean in Stata syntax?
Marks a variable as continuous (numeric).
What does i. mean in Stata syntax?
Marks a variable as categorical (indicator).
In linear regression, how is the total effect for a category calculated?
Add independent exposure effect + interaction effect.
In logistic regression, how is the total effect calculated?
Multiply the independent exposure OR by the interaction term’s OR.
What does a non-significant interaction term mean?
Effect of exposure on outcome does not differ across levels of the modifier.
What is the total effect in linear regression?
The sum of the independent exposure effect and the interaction effect.
What indicates no effect modification after stratification?
Similar effect estimates across strata
What does a significant p-value (p<0.05) for interaction mean?
Presence of statistically significant effect modification.
When should interactions be tested?
At final stage of multivariable analysis, focusing on pre-specified, biologically plausible ones.
What is the preferred order of notation?
Interacting variable first: X_variable##E_variable.
What does the coefficient of an interaction term represent?
Difference in exposure effect on outcome between categories of the modifier.
-
Section 1: Introduction to Measurement: types of variables and types of distributions67
-
Section 2 - Introduction to Statistical Analysis for Epidemiology & PH60
-
Section 4: descriptive analysis of numeric data64
-
Section 5: Descriptive Analysis of Categorical Data32
-
Section 6: Basic analysis for Determining Associations with numeric outcomes I: T-test and ANOVA26
-
Section 7: basic analysis for determining associations with numeric outcomes II: Scatterplots & correlation analysis34
-
Section 8: Analysis of two categorica variables: a categorical independent - exposure and a categorical dependent - outcome35
-
Section 9: Non-parametric Statistical Tests45
-
Section 10: Linear Regression Analysis140
-
Section 11: Multiple Linear Regression Analysis107
-
Section 12: Univariate and Multivariate Logistic Regression analysis59
-
Section 13: Assessing Interactions using Linear and Logistic Regression analysis23
-
Section 14: Assessing confounding and mediations using Linear and Logistic regression analysis32
-
Section 15: Survival (time-to-event) analysis31
-
Section 3: Intro to stata33
