Sample Stata File

Stata’s .dta is the native dataset format of the Stata statistical package, ubiquitous in economics, epidemiology, and survey research. This sample uses format version 118 — the dialect introduced with Stata 14 — which is the one to standardize on: it stores strings as UTF-8 (older versions were effectively Latin-1) and supports large observation counts. It was written by pandas.DataFrame.to_stata(..., version=118).

The format is binary and self-describing: a header carries the observation count and variable metadata, followed by fixed-layout data records. Two Stata-specific conventions show up even in a file this small:

• Variable name rules. Stata names are limited to 32 characters, may contain only letters, digits, and underscores, and cannot start with a digit. The employee headers (name, start_date, …) already comply, so no renaming occurred — but a CSV header like 2024 salary or first-name would be rewritten on export, which is a common source of column-name drift between Stata and everything else.
• Typed columns. salary is stored as a Stata long (32-bit integer), and each string column is a fixed-width str# type sized to its longest value. Unlike CSV, the types travel with the file. start_date is kept as an ISO 8601 string rather than a Stata internal date (days since 1960-01-01) so it reads identically in pandas, R, and Stata without format directives.

What this file omits is also informative: real research .dta files usually carry variable labels, value labels (integer codes mapped to text), and missing-value sentinels (., .a–.z). This sample has none, making it a clean baseline before testing those features with your own data.

To open it, use pandas.read_stata(), R’s haven::read_dta(), or Stata 14+ — or follow the DTA to CSV guide for every free conversion route, including no-code options. The Parquet and Feather siblings have in-browser converters if you need CSV output today.