Behavioral sciences 

 
Quantitative behavioral scientists rely on Stata for its accuracy, extensibility, reproducibility, and ease of use. Whether you are researching social trends, cognitive neuroscience, learning, or psychometric methods, Stata provides all the statistics, graphics, and data-management tools needed to pursue a broad range of behavioral science questions.



Features for behavioral scientists

Item response models
Use item response models to reveal unobservable characteristics using questionnaires. Obtain parameter estimates and graphs from binary response models, ordinal response models, categorical response models, or a mixture. And more.

 

Structural equation models
Construct models for continuous, binary, count, ordinal, multinomial, or survival outcomes and incorporate unobserved components at any level. Specify weights at each level of the model. Use with complex survey data. And more.

 

Survey methods
Handle probability sampling weights, multiple stages of cluster sampling, stage-level sampling weights, stratification, and poststratification. Use variance techniques of balanced repeated replications, the bootstrap, the jackknife, successive difference replication, and linearization. Fit many different statistical models on complex survey data. And more.

 

Linear and generalized linear models (GLMs)
Fit linear, quantile, truncated, and censored regressions and maximum likelihood models for binary, count, fractional, continuous, ordered, and multivariate outcomes. And more.

 

ANOVA/MANOVA
Use ANOVA and multivariate ANOVA to test for differences between continuous outcomes by groups. Study balanced, unbalanced, factorial, nested, and mixed designs. Analyze repeated measures, marginal means, and contrasts. Draw profile and interactions plots. And more.

 

Contrasts, pairwise comparisons, and margins
Use estimation results to obtain estimates and graphs of interactions, average effects, partial effects, contrasts, and pairwise comparisions. Draw profile and interactions plots. And more.

 

Treatment effects
Estimate treatment effects for continuous, binary, count, and survival outcomes and for multilevel and multivalued treatments. Obtain estimates of effects under endogeneity. Choose from inverse probability weights (IPW), propensity-score matching, covariate matching, regression adjustment, doubly-robust augmented IPW and IPW with regression adjustment models. And more.

 

Hierarchical models
Fit fixed-effects, random-effects, and mixed-effects models to multilevel data with continuous, binary, and count outcomes. Construct models for different correlation structures and nesting levels. And more.

 

Multiple imputation
Use descriptive statistics such as means, proportions, and ratios, fit linear and nonlinear regressions, multilevel mixed-effects models, panel-data models, survival models, and much more using multiple imputation to account for missing data in your sample.

 

Power and sample size
Determine the sample size needed for your experiment to recover meaningful effects without wasting resources. Obtain one-sample and two-sample tests of means, variances, proportions, and correlations. And more.

Bayesian analysis
Fit Bayesian regression models using a Metropolis–Hastings Markov chain Monte Carlo (MCMC) method. Choose from a variety of supported models or program your own. Check convergence visually using extensive graphical tools. Compute posterior mean estimates and credible intervals for model parameters and functions of model parameters. Perform interval and model-based hypothesis testing. Compare models using Bayes factors.

 

Mata
Program your own estimator using Stata's built-in matrix language MATA. Use MATA interactively with Stata. Obtain inversions, decompositions, eigenvalues and eigenvectors, and numerical derivatives. Use LAPACK routines, real and complex numbers, string matrices, and object-oriented programming. And more.