Review set
Reading inverse-probability weights
The questions ask you to form weights and to say what their distribution reveals about whether a comparison is supported. Forming a weight comes back readily; reading a weight of 400 as a limit of the evidence rather than a computational annoyance is what needs unlearning.
What this involves
Retrieval after a delay, which is what makes knowledge durable.
Answers here count towards what the system knows about your skills.
What this covers
- Compute and diagnose a weighted estimate
Given estimated propensity scores, the learner can form inverse-probability weights, compute the weighted estimate, and diagnose what extreme weights indicate about the comparison.