Home  /  Products  /  Features  /  Causal inference/Treatment effects
Order

Causal inference/Treatment effects

Stata's causal-inference suite allows you to estimate experimental-type causal effects from observational data. Whether you are interested in a continuous, binary, count, fractional, or survival outcome; whether you are modeling the outcome process or treatment process; Stata can estimate your treatment effect. With the most comprehensive set of causal-inference estimators available in any software package, you will find the one that's right for you.

Learn about causal inference and causal-inference analysis.

See Difference-in-differences (DID) and DDD models.

See Heterogeneous difference in differences (DID).

See Treatment-effects lasso estimation.

Estimators

Endogeneity, Heckman-style selection, and panel data with causal effects

  • Linear regression
  • Interval regression, including tobit
  • Probit regression
  • Ordered probit regression
  • Exogenous or endogenous regressors
  • Endogenous or exogenous treatment; binary or ordinal treatment
  • Random-effects models for panel data

Statistics

  • Average treatment effects (ATEs)
  • ATEs on the treated (ATETs)
  • Potential-outcome means (POMs)

Outcomes

  • Continuous—linear
  • Binary—logistic, probit, heteroskedastic probit
  • Count—Poisson
  • Fractional
  • Nonnegative, including exponential mean
  • Survival—exponential, Weibull, gamma, lognormal

Treatments

  • Binary—logistic, probit, heteroskedastic probit
  • Multivalued-multinomial logistic

Diagnostics

Postestimation Selector

  • View and run all postestimation features for your command
  • Automatically updated as estimation commands are run

Causal mediation analysis New

  • Continuous, binary, and count outcomes
  • Continuous, binary, and count mediators
  • Binary, multivalued, and continuous treatments
  • Linear, logit, probit, Poisson, and exponential mean models
  • Direct effects, indirect effects, total effects, and POMs

Endogenous treatment effects

Difference-in-differences (DID) and triple-differences (DDD)
estimation

  • DID and DDD estimators for repeated cross-sections data
  • DID and DDD estimators for panel data
  • DID diagnostics and tests
    • Test and graphs for parallel trends
    • Granger causality test
    • Time-specific treatment effects
  • ATET inference with small number of treatment and
    control groups
    • Bacon decomposition New
    • Wild bootstrap
    • Donald–Lang estimator
    • Bias-corrected cluster–robust SEs
    • Bell–McCaffrey degrees of freedom

Heterogeneous DID

  • Four estimators
    • regression adjustment (RA)
    • inverse probability weighting (IPW)
    • augmented inverse probability weighting (AIPW)
    • two-way fixed-effects regression (TWFE)
  • Estimation of heterogeneous treatment effects
    • Panel data
    • Repeated cross-sectional data
  • Graphical representation of treatment effects
  • Estimate and visualize aggregations of ATETs within
    • cohort
    • time
    • exposure to treatment
  • Simultaneous confidence intervals

Treatment effects with high-dimensional controls

  • Continuous, binary, and count outcomes
  • Logit or probit treatment model
  • ATEs, ATETs, and POMs
  • Lasso or square-root lasso variable selection
  • Neyman orthogonal and doubly robust estimator
  • Double machine learning
  • Flexible model specification

Additional resources

dialog box for teffects
Watch Tour of treatment-effects estimators in Stata.

Watch Tour of treatment-effects estimators in Stata.
Watch Introduction to treatment effects, part 1.
Watch Introduction to treatment effects, part 2.

See New in Stata 18 to learn about what was added in Stata 18.