Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →asgard-ai-platform/stat-causal-inference
Apply causal inference methods — counterfactual framework, instrumental variables, propensity score matching, and difference-in-differences — to estimate causal effects from observational data.
Passed 23 of 25 test cases with the skill, and 22 of 25 without it.
1 fixed22 passed either way2 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
A test case the skill fixed. Same model, same prompt, with and without the skill.
Our hospital system offered a specialized outpatient follow-up program to cardiology patients upon discharge, but enrollment was voluntary. We have access to extensive electronic health records covering patient baseline health and demographic variables prior to discharge. We need to determine if participating in the program causally reduces 30-day readmission rates. Please supply a complete causal assessment document structured with: a Causal Question section (treatment, outcome, counterfactual), an Identification Strategy section (method justification, underlying assumption, assumption verification), a Results section (point estimate with confidence interval and robustness checks), and a Limitations section detailing potential unobserved confounding or bias.
| Checked | Without skill | With skill |
|---|---|---|
| Selects Propensity Score Matching as the identification methodology. | Missed | Passed |
| States the assumption of selection on observables. | Passed | Passed |
| Warns that voluntary enrollment introduces unobserved confounding. | Passed | Passed |
The instructions your model reads when the skill loads.
IRON LAW: Correlation Is Not Causation — But Causation Is Estimable
Observational data cannot prove causation through correlation alone.
BUT with the right methodology (matching, IV, DID, RDD), we CAN
estimate causal effects from observational data — IF the assumptions
of each method are satisfied and explicitly tested.
The key question is always: "What would have happened WITHOUT the treatment?"
(the counterfactual)
We observe: Y_i(treated) — what happened to the treated unit.
We want to know: Y_i(treated) - Y_i(untreated) — the causal effect.
We can never observe: Y_i(untreated) for the same unit at the same time.
All causal inference methods estimate the counterfactual — what would have happened without the treatment.
| Method | When to Use | Key Assumption |
|---|---|---|
| RCT | You can randomize | Random assignment eliminates confounders |
| Propensity Score Matching (PSM) | Treatment is non-random but based on observables | No unobserved confounders (selection on observables) |
| Instrumental Variables (IV) | Unobserved confounders exist but you have an instrument | Instrument affects treatment but not outcome directly |
| Difference-in-Differences (DID) | Policy/event creates natural treatment/control groups | Parallel trends: groups would have trended similarly without treatment |
| Regression Discontinuity (RDD) | Treatment assigned by a cutoff | Observations just above/below cutoff are comparable |
| Synthetic Control | One treated unit, multiple control units (aggregate data) | Synthetic weighted combination matches pre-treatment trends |
test cases passed, vs 22 / 25 without
Published skills whose descriptions closely match this one, with their headline benchmark lift.
More skills from the same source and nearby in the registry, with their headline benchmark lift.