Setup | Basic workflow | Step 1: Fit a propensity score model | Step 2: Calculate weights and fit a weighted outcome model | Step 3: Estimate causal effects | Choosing an estimand | Handling extreme weights | Overlap estimands | Trimming | Truncation | Which approach? | Interpreting results | Binary outcomes | Continuous outcomes | Next steps | Continuous exposures | Categorical exposures | Calibration | Censoring weights | Learning more

New ornaments | Ornament factories | Shafts | Line types | Variable width | Distortion | Scales

Head ornaments | Wings | Line | Minimal | Fins ornaments | Feather | Other ornaments | Cups

Aesthetics | Heads, fins and middle arrows | Using arrow parts | Setting sizes | Middle arrows | Resection | Other parameters | justify | force_arrow | sep

Setup | Running check_positivity() | Reading the container | PoRT: locating the subgroups | Effective data points: how much support surrounds each intervention | Extrapolation: distance and the convex hull | Choosing diagnostics beyond the defaults | References

Setup | Effective data points: support at chosen exposure values | Shifting each observation away from the dose it received | Hat values: leverage-based extrapolation risk | Highest density region non-overlap | When each diagnostic is informative | Declaring the exposure type | Time-varying exposures | Where to go next | References

Directed Acyclic Graphs | Structural Causal Graphs | Relationships between variables | Colliders and collider-stratification bias | Mediation | Resources

Introduction | Confounders and confounding | Colliders, M-bias, and butterfly bias | Measurement error | Selection bias | Resources

Two families of methods | Which exposures and estimands each method supports | Exact versus tolerance-based balance | Constraints in depth | Continuous exposures | A note on solvers

Why the weights' uncertainty matters | M-estimation with ipw() | Categorical exposures | Effect modification with .by | Joint exposures | Continuous exposures | Covariate-adjusted outcome models | Bootstrapping the other fits | Choosing an approach

Two different questions | A dataset where they diverge | Effect modification with .by | Interaction with a joint intervention | Why the two numbers differ | Interaction is symmetric | Odds ratios and effect modification | Guardrails on joint weights | Two notes on labels | Learning more

Setup | ETA.Bias: the bias an estimator incurs | A categorical exposure, term by term | A continuous exposure and its working model | Extrapolation as the g-computation view of positivity | Where to go next | References

Setup | What the summaries measure | A hand-built point-treatment example | A mean below one signals leaked support | Time-varying treatments and cumulative products | A fitted longitudinal policy with lmtp | Inverse-probability weights from a propensity model | Where to go next | References

Setup | Finding subgroups with PoRT | Tuning the reading rule | Continuous exposures | Sequential violations with sPoRT | Limiting the history with lag | Censoring | Where to go next | References

Overview | Fitting your first model | Inspecting the fit | Other regression models | The estimator objects | Stacking estimating equations | Causal inference: inverse probability weighting | Predictions | Clustered data | Exact differentiation | Next steps

Overview | Linear regression | Logistic regression | Poisson regression | GLM: generalized linear models | Penalized regression | Ridge regression | LASSO regression | Elastic net | Robust regression | Weighted regression | Predictions after regression

Overview | Simulated data | G-formula (g-computation) | Inverse probability weighting (IPW) | Truncating propensity scores | Augmented IPW (doubly robust) | Comparing methods | Further reading

A confounded example | Fitting balancing weights | Checking balance with halfmoon | Estimating a causal effect | Standard errors with ipw() | Where to go next

Overview | Writing a custom estimating equation | Example: estimating a ratio of means | Key points for custom equations | Stacking custom EEs with built-in ones | Example: log odds-ratio from logistic regression | How stacking works | Using the delta method | Example: odds ratio via the delta method | When to use the delta method vs. stacking | Using the delta method with raw estimates

Overview | Creating Directed Acyclic Graphs | Working with DAGs | Plotting DAGs | Plotting directly in ggplot2

Setup | Analysis | Analysis incorporating time varying data