Stata orientation and the do-file workflow

Stata is the working tool of academic economists, policy researchers, and the World Bank / IMF / central bank research departments. It is purpose-built for the kinds of analyses applied economists actually do — panel regressions, IV, fixed effects, survey data — with a stable, well-documented command syntax that hasn't changed much in 30 years.

Why Stata, in a Python world

Python and R have eaten most of the data-science world. Stata is the place that survives because: (1) it has by far the most-cited econometric routines (xtreg, ivreg2, reghdfe, esttab — all written by economists, for economists); (2) the help documentation is extraordinarily good; (3) the file format and command grammar haven't changed, so a do-file from 2005 still runs today; (4) regulators, ministries, and many academic journals expect Stata-compatible reproduction packages.

The four windows

• Command — type one command at a time, see the result
• Results — what's printed back
• Variables — the columns of the loaded dataset
• Properties — metadata on the selected variable

Always work in do-files

Typing commands into the Command window is fine for exploration but useless for serious work. A do-file is a text file (with .do extension) containing your commands. You execute it with `do filename.do`, and the analysis is reproducible. Every published paper that uses Stata ships a do-file.

* my_analysis.do

set more off                      // don't pause output
capture log close                 // close any open log
log using analysis.log, replace   // start a new log

use bankrates.dta, clear
summarize

log close

Working directory

pwd                          // print working directory
cd "C:/projects/banking"      // change directory

The help system

help command-name opens the manual entry for that command. Stata's manuals are extraordinarily good — better than R's, better than Python's. When in doubt, help [whatever] is the first move.