What the results depend on | Why this works | Reproducing results from earlier versions | The simple argument | BCa confidence intervals | Choosing a parallel backend | Using futurize | Progress bars | A note on statistic

Introduction | Identifying the estimand | G-computation | Modeling the Outcome | Estimating Standard Errors and Confidence Intervals | Asymptotically Correct Standard Errors Using M-estimation | Robust Standard Errors | Bootstrapping | Estimating Treatment Effects and Standard Errors After Weighting | The Standard Case: Binary Treatment with Asymptotically Correct SEs | Adjustments to the Standard Case | Weighting for the ATT or ATC | Weighting for estimands other than the ATT, ATC, or ATE | Binary, count, and categorical outcomes | Survival outcomes | Using sampling weights with complex surveys | Multi-Category Treatments | Continuous Treatments | Longitudinal Treatments | Moderation Analysis | Reporting Results | References | Code to Generate Data used in Examples

Introduction | Marginal Structural Models for Longitudinal Treatments | The Estimand | Time-Varying Confounding | Assumptions | Inverse Probability Weights | Estimating Weights for a Longitudinal Treatment | The Data | Initial Imbalance | Estimating the Weights | Assessing Balance | Estimating the Treatment Effect | Accounting for Loss to Follow-Up | The Estimand with Censoring | Simulating Loss to Follow-Up | Other Considerations | References

Propensity score weighting using GLMs (method = "glm") | missing = "saem" | Binary and multi-category treatments with link = "br.logit" | Multi-category treatments with multi.method = "mclogit" | Multilevel treatment models with random effects | Multi-category treatments with multi.method = "mnp" | Propensity score weighting using GBM (method = "gbm") | Covariate Balancing Propensity Score weighting (method = "cbps") | Nonparametric Covariate Balancing Propensity Score weighting (method = "npcbps") | Entropy balancing (method = "ebal") | Inverse probability tilting (method = "ipt") | Stable Balancing Weights (method = "optweight") | Propensity score weighting using SuperLearner (method = "super") | Propensity score weighting using BART (method = "bart") | Energy balancing (method = "energy") | Characteristic function distance balancing (method = "cfd")

Introduction | Balancing Weights for a Point Treatment | Balancing Weights for a Longitudinal Treatment | Accounting for censoring | References

Introduction | Weighting | Balance and Effective Sample Size | Estimands and Target Populations | Treatment Types | Parametric Methods | Propensity Score Weighting Using Generalized Linear Models (method = "glm") | Covariate Balancing Propensity Score Weighting (method = "cbps") | Inverse Probability Tilting (method = "ipt") | Machine Learning Methods | Propensity Score Weighting Using Generalized Boosted Models (method = "gbm") | Propensity Score Weighting Using SuperLearner (method = "super") | Propensity Score Weighting Using BART (method = "bart") | Optimization-Based Methods | Entropy Balancing (method = "ebal") | Stable Balancing Weights (method = "optweight") | Nonparametric Covariate Balancing Propensity Score Weighting (method = "npcbps") | Energy Balancing (method = "energy") | Characteristic Function Distance Balancing (method = "cfd") | Additional Options | Balancing Higher Moments, Interactions, and Quantiles (moments, int, quantile) | Stabilizing Weights (stabilize) | Trimming Weights (trim()) | Calibrating Propensity Scores (calibrate()) | Marginal Mean Weighting Through Stratification (subclass) | Supplying Propensity Scores (ps) | Choosing a Weighting Method | Summary of the Weighting Methods | A Workflow for Selecting a Method | Reporting the Weighting Specification | References

Introduction | Recommendations for Balance Assessment | Recommendations for Balance Reporting | Assessing Balance with MatchIt | summary.matchit() | plot.summary.matchit() | plot.matchit() | Assessing Balance After Subclassification | Assessing Balance with cobalt | bal.tab() | love.plot() | bal.plot() | Conclusion | References

Introduction | Identifying the estimand | G-computation | Modeling the Outcome | Estimating Standard Errors and Confidence Intervals | Robust and Cluster-Robust Standard Errors | Bootstrapping | Estimating Treatment Effects and Standard Errors After Matching | The Standard Case | Adjustments to the Standard Case | Matching for the ATE | Matching with replacement | Matching without pairing | Propensity score subclassification | Binary outcomes | Survival outcomes | Using Bootstrapping to Estimate Confidence Intervals | The standard bootstrap | The cluster bootstrap | Moderation Analysis | Reporting Results | Common Mistakes | 1. Failing to include weights | 2. Failing to use robust or cluster-robust standard errors | 3. Interpreting conditional effects as marginal effects | References | Code to Generate Data used in Examples

Introduction | Matching | Nearest Neighbor Matching (method = "nearest") | Optimal Pair Matching (method = "optimal") | Optimal Full Matching (method = "full") | Generalized Full Matching (method = "quick") | Genetic Matching (method = "genetic") | Exact Matching (method = "exact") | Coarsened Exact Matching (method = "cem") | Subclassification (method = "subclass") | Cardinality and Profile Matching (method = "cardinality") | Customizing the Matching Specification | Specifying the propensity score or other distance measure (distance) | Implementing common support restrictions (discard) | Caliper matching (caliper) | Mahalanobis distance matching (mahvars) | Exact matching (exact) | Anti-exact matching (antiexact) | Matching with replacement (replace) | $k$:1 matching (ratio) | Matching order (m.order) | Choosing a Matching Method | Reporting the Matching Specification | References

Introduction | Matching | Assessing Balance | Estimating the Effect | Code to Generate Data used in Examples | References

Introduction | Planning | Check Initial Imbalance | Matching | Assessing the Quality of Matches | Trying a Different Matching Specification | Estimating the Treatment Effect | Reporting Results | Conclusion | References

How are standardized mean differences computed in cobalt? | Why are mean differences not standardized for binary covariates? | Why do you use the same standardization factor before and after adjustment? | How do I extract the balance tables from the bal.tab() object? | How are balance statistics computed when using subclassification? | Why don't I get the same balance statistics when using cobalt as I do when using tableone? | Why doesn't thresholds work with bal.tab() with multiply imputed or clustered data? | References

Introduction | Citing cobalt | Why cobalt? | Visual clarity | Useful summaries | One tool to rule them all | Flexibility | Pretty plots | Unique features | How To Use cobalt | Utilities | bal.tab() | Using bal.tab() on its own | Using bal.tab() with MatchIt | Using bal.tab() with WeightIt | bal.plot() | love.plot() | Additional Features | Using cobalt with continuous treatments | Using cobalt with multi-category treatments | Using cobalt with censoring | Comparing balancing methods | Using the prognostic score for balance assessment | Details on Calculations | Variance in Standardized Mean Differences and Correlations | Weighted Variance | Effective Sample Size for Weighting | What's Missing in cobalt | Test Statistics and P-values | Q-Q Plots and Summaries | What's Added in cobalt | Density Plots | Variance Ratios | Distinguishing Continuous and Binary Covariates | Interactions and Polynomials | Clusters | Missing Data | For Programmers: Integrating cobalt with Your Package | Acknowledgments | References

References

Setup | bal.tab() | bal.plot() | love.plot() | Censoring | Other Packages | References

bal.compute() and bal.init() | Balance statistics | smd.mean, smd.max, smd.rms | ks.mean, ks.max, ks.rms | ovl.mean, ovl.max, ovl.rms | mahalanobis | energy.dist | kernel.dist | l1.med | r2, r2.2, r2.3 | p.mean, p.max, p.rms | s.mean, s.max, s.rms | distance.cov, distance.cor | Choosing a balance statistic | Example | Tuning GBM for balance | References

cobalt and Segmented Data | Clustered Data | bal.tab() | bal.plot() | love.plot() | Multiply Imputed Data | Multi-Category Treatments with Multiply Imputed Data | Concluding Remarks

Introduction | Effect Curves | This Guide | The Data | Analysis | Fitting the outcome model | Computing and plotting the ADRF | Testing the ADRF | Estimating a linear projection | Other effect curves | Reference effect curves | The AMEF | Subgroup ADRFs | Male = 0 | Male = 1 | Subgroup ADRF contrasts | Additional Scenarios | Nonlinear outcome models | Transformed | Untransformed | Untransformed | Balancing weights | Sampling weights | Using survey | Using glm() | Sampling weights and balancing weights together | Other scenarios | Multiply imputed data | Using MatchThem | Bayesian models | References

Utilities | splitfactor() and unsplitfactor() | get.w() | bal.tab() | Using bal.tab() with twang | Using bal.tab() with Matching | Using bal.tab() with optmatch | Using bal.tab() with CBPS | Using bal.tab() with ebal | Using bal.tab() with designmatch | Using bal.tab() with sbw | Using bal.tab() with MatchThem | Using bal.tab() with cem | Using bal.tab() with other packages

Introduction | Related software | Using clarify | 1. Fitting the model | 2. Drawing from the coefficient distribution | 3. Computing derived quantities | 4. Summarize and visualize the simulated distribution | Wrappers for sim_apply(): sim_setx(), sim_ame(), and sim_adrf() | sim_setx(): predictions at representative values | sim_ame(): average adjusted predictions and average marginal effects | sim_adrf(): average dose-response functions | Transforming and combining estimates | transform() | cbind() | Using clarify with multiply imputed data | Comparison to other packages | Conclusion | References

Introduction | Predictions at representative values | Zelig workflow | clarify workflow | Rare-events logit | Estimating the ATT after matching | Combining results after multiple imputation | References