Stats Console
BetaStata-compatible statistics in your browser: open a .dta file, fit models, draw graphs, write do-files. Free, with no sign-up.
wages.dta500 obs · 11 vars
Stats Console beta. Stata-compatible commands, run in this browser tab.
Type help for the command list. Tab completes commands and variable names; Up and Down recall commands.
. sysuse wages
(Synthetic wage data, Folio example)
. summarize wage educ exper hours
Variable | Obs Mean Std. dev. Min Max
-------------+---------------------------------------------------------
wage | 500 11.85866 4.966043 2.74 33.81
educ | 500 12.938 2.427745 8 20
exper | 500 22.242 12.60497 0 47
hours | 483 37.80745 6.117639 18 54
. regress wage educ exper i.region, vce(robust)
Linear regression Number of obs = 500
F(5, 494) = 25.35
Prob > F = 0.0000
R-squared = 0.1924
Root MSE = 4.4854
------------------------------------------------------------------------------
| Robust
wage | Coefficient std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
educ | .8344353 .0826583 10.10 0.000 .6720301 .9968404
exper | .0844597 .014982 5.64 0.000 .0550234 .1138959
|
region |
Midwest | -.4982768 .542929 -0.92 0.359 -1.565012 .568458
South | -.8081702 .5723987 -1.41 0.159 -1.932806 .3164661
West | .6052676 .5913103 1.02 0.307 -.5565256 1.767061
|
_cons | -.6065513 1.284379 -0.47 0.637 -3.130071 1.916968
------------------------------------------------------------------------------
Beta: results are still being compared with Stata’s own output, so confirm key figures before you report them. Stata is a registered trademark of StataCorp LLC; Stats Console is an independent implementation, not affiliated with or endorsed by StataCorp.
Checked against known answers
Each row is an automated test in Folio’s test suite. Choose Examples, then Longley benchmark, to rerun the first one yourself.
| Area | What the test confirms |
|---|---|
| Linear regressionNIST StRD Longley benchmark | Every certified coefficient, standard error and analysis-of-variance quantity, to eight or more significant digits. |
| Robust and clustered errorsClosed-form sandwich estimators | vce(robust) is HC1; vce(cluster) applies G/(G−1)·(N−1)/(N−k) with G−1 degrees of freedom. |
| Logit and probitClosed forms for a binary regressor | The log odds ratio and its textbook standard error; probit coefficients equal the normal quantiles of the cell proportions. |
| t tests and ANOVATextbook formulas | One-sample, pooled, Satterthwaite and paired t tests; the one-way F equals the regression F on group dummies; Bartlett’s test. |
| Panel modelsExact identities | Fixed effects equal the dummy-variable regression; random effects equal OLS on θ-transformed data; between effects equal OLS on panel means. |
| PostestimationExact identities | A one-coefficient test is the squared t; margins and dydx() equal averages of predictions computed separately, with delta-method standard errors; hausman matches its closed form. |
| Worked examplesOutput printed in Stata’s manuals | ranksum, signrank, ttest and ologit reproduce the manuals’ worked examples line for line, run on the data those examples print. tabulate and tabi reproduce their tables and statistics, without the exact test’s enumeration log; poisson reproduces the results tables of its two examples. codebook, misstable and xtdescribe reproduce selected printed tables on data built to the same counts. |
| Exact testsFull enumeration | Exact p-values of ranksum, signrank and Fisher’s exact test equal sums over every rank assignment, or every table with the same margins. |
| Count and ordered modelsLikelihoods maximized separately | poisson, nbreg, ologit and oprobit agree with likelihoods coded and maximized in the test; ordered models of a binary outcome equal logit and probit. |
| Instrumental variablesClosed forms and the two-step method | 2SLS equals two-step regression with structural residuals; just-identified fits equal their closed forms; first-stage, Durbin–Wu–Hausman and Sargan statistics match direct computations. |
Not yet checked: commands beyond these worked examples against Stata’s own output, and .dta files written here opened in Stata.
Commands
Stata syntax, abbreviations included. Type help in the console for the full list.
- Look at data
describecodebooklistbrowsesummarizetabstattabulatecorrelatepwcorrmisstablextsumxtdescribeci means- Test
ttestttestitabionewayranksumsignrankkwallisspearmanExact p-values for ranksum and signrank; Fisher’s exact test in tabulate.- Estimate
regresslogitlogisticprobitpoissonnbregologitoprobitivregressxtregRobust and clustered errors; i. and ## factor variables; xtreg with re, fe and be.- After estimation
predicttesttestparmmarginsmarginsplothausmanestatestimates tableetableestat runs firststage, endogenous and overid after ivregress; etable exports to LaTeX, HTML, Markdown and plain text.- Graph
histogramscattertwowaygraph bargraph boxDownload any graph as PNG or SVG.- Change data
usesavemergeappendreshapeimport delimitedexport delimitedgeneratereplaceegenrecodeencodecollapseformatduplicatesassert- Program
localglobaldisplayforeachforvalueswhileifbyquietlycapture
Keyboard
- Enter
- Run the command
- ↑↓
- Recall earlier commands
- Tab
- Complete a command or variable name
- Esc
- Cancel an open loop or block
- CtrlEnter
- Run the do-file, or the selected lines
Questions
- Is this Stata?
- No. It is an independent console that reads Stata’s command syntax and prints results in Stata’s layout. It covers the commands listed on this page, not every command or option, and it is in beta. Stata is a registered trademark of StataCorp LLC; Folio is not affiliated with or endorsed by StataCorp.
- Is it free?
- Yes, free with no sign-up. Everything runs in your browser.
- Is my data uploaded?
- No. A file you open is read inside the browser tab, and every command runs there too. Closing the tab clears the dataset unless you save it as a file; the do-file editor keeps its text in this browser only.
- Can I open my .dta files?
- Yes. It reads .dta files from Stata 8 onward (formats 113 to 119) and CSV files, and the save command writes format 118, the format Stata 14 and later use.
- How accurate are the results?
- The regress command reproduces every certified value of the NIST StRD Longley benchmark to eight or more significant digits. Several commands, among them ranksum, ttest and ologit, reproduce worked examples from Stata’s manuals line for line, and the test suite checks the others against exact identities, such as fixed effects against the regression with one dummy per panel. Not every command has been compared with Stata’s own output yet, so confirm key results for a thesis or paper in your usual package.