Chapter 5 (5.5-5.7) - Internal vs. External Validity Flashcards
What does internal validity explore?
If the estimated causal effect is valid for the sample of observations in a given study; the answer depends on whether or not you’ve successfully eliminated all potential confounders.
What does external validity explore?
If the estimated causal effect is valid beyond a given study; the answer depends on 1. if the sample of observations is representative of the population in the study and 2. if the treatment used in the study is representative of the treatment for which you want to generalize the results.
True or false: in observational studies, you risk failing to account for all potential confounders.
True
True or false: in general, studies based on randomized experiments tend to have weak internal validity, but strong external validity.
False
True or false: in general, observational studies tend to have weak internal validity, but strong external validity.
True
What’s the best way to ensure that a sample is representative of the population?
By using random sampling.
What’s the best way to ensure that the treatment group and the control group are comparable?
By using random treatment assignment.
True or false: you can rely on random treatment assignment to eliminate potential confounders.
False
What does internal validity “ask”?
“Is the estimated causal effect valid for the sample of observations in the study?”.
What does external validity “ask”?
“Is the estimated causal effect valid beyond this particular study?”.
Regarding what internal validity “asks”, what does the answer depend on?
If the treatment group and the control group are comparable.
Regarding what external validity “asks”, what does the answer depend on?
If the sample of observations in the study is representative of the population.
What does randomization ensure?
- that the sample is representative of the target population
- strong external validity
What does random treatment assignment do?
- eliminates all potential confounders (making the treatment group and the control group comparable)
- ensures strong internal validity