StataCorp Stata

About StataCorp Stata

Stata is the environment econometrics, epidemiology and the social sciences settle on when the analysis has to be reproducible by somebody else. Everything happens through commands, every command is logged, and a do-file rerun on another machine produces the same numbers. That property is the whole reason it survives against tools with prettier interfaces.

The estimation set is enormous and consistently structured. Linear and generalised linear models, panel data with fixed and random effects, survival analysis, multilevel mixed models, structural equation modelling, instrumental variables, time series, survey weighting and multiple imputation all share the same syntax pattern, so learning one command teaches the shape of the rest. Post-estimation commands attach to whatever was just fitted rather than needing the model rebuilt.

Data management is not an afterthought here. Reshaping long to wide, merging on keys with a match report, collapsing to summaries, handling value labels and dealing with string encoding are all first class operations, and the variables manager keeps a large dataset legible while it is being cleaned.

Output is built for papers rather than screenshots. Tables export to publication formats, graphs are fully controllable down to the axis tick and save as vector files, and the whole session can be captured to a log that becomes the appendix. Community written commands install into the same namespace as the built-in ones.

What is included in this build

• Full estimation set including panel, survival, multilevel, structural equation and time series models
• Consistent command syntax across the entire estimation library
• Post-estimation commands for margins, predictions, tests and diagnostics
• Data management with reshape, merge, collapse, labels and Unicode string handling
• Do-file scripting with complete session logging for reproducible analysis
• Publication ready table export and fully controllable vector graphics
• Matrix programming language for custom estimators
• Community command installation into the same namespace as built-in commands

Changed in version 17.0

• Faster handling of very wide datasets during reshape operations
• Improved graph rendering on high resolution displays
• Additional post-estimation diagnostics exposed on mixed models
• Better memory behaviour on datasets above ten million observations
• Corrected a rounding issue in one of the survey variance estimators

Installation walkthrough

These steps assume a clean machine. If a previous release of the same application is present, deal with that first, since an overlapping install is the most common cause of a setup that stops halfway.

1. Disable real-time protection until the installation has completed.
2. Extract the archive to a short folder path with no spaces.
3. Run setup as administrator and accept the default program folder.
4. Choose the edition tier during setup, since it sets the observation and variable ceilings.
5. Launch once and set the working directory and default memory allocation before loading real data.

System requirements and file details

Frequently asked about this title