Stata Cheat Sheet

The core commands for this course

A short reference for the Stata commands we actually use in PAI 721. Every command has more options than shown here; type help followed by any command name (for example, help summarize) to see them all. Words like var, group, y, and x are placeholders for your own variable names.

Don’t forget - for a more extensive references, you also have:

• Solutions to every problem set (course drive) and code embedded in class slides
• The UCLA Stata guide, linked in the syllabus
• The Acock “Gentle Introduction to Stata” optional/recommended reference book

Setup & Files

Command	What it does
cd "folder"	Set the working directory (where Stata looks for files)
use "file.dta", clear	Open a Stata dataset, closing whatever is already open
import delimited "file.csv", clear	Open a .csv file
save "file.dta", replace	Save the data in memory, overwriting the old copy
clear	Wipe the data in memory
browse	Open the spreadsheet view of your data
describe	List the variables in the dataset

Note.do files

A .do file is a plain-text list of commands that Stata runs top to bottom. Open the editor with doedit. Lines starting with * or // are comments; use them to explain what you’re doing and to label which question each chunk of code answers. Put /// at the end of a line to continue a long command onto the next line.

Looking at Data

Command	What it does
sum var	Mean, standard deviation, min, max, and n (summarize)
sum var, detail	Adds the median and percentiles (25th = Q1, 75th = Q3)
tab var	Frequency table: how many cases take each value
tab var1 var2	Cross-tabulation of two variables
tabstat var, stats(mean sd n)	Choose your own summary statistics
tabstat var, by(group)	The same, separately for each group
bys group: sum var	Run any command separately for each group
corr var1 var2	Correlation coefficient
sort var	Sort cases from low to high (gsort -var for high to low)

Creating & Changing Variables

Command	What it does
gen new = pts / g	Create a new variable from a formula
replace var = 0 if var == 9	Change values that meet a condition
recode var 7=., gen(newvar)	Recode values into a new variable (here, 7 becomes missing)
keep if year == 2010	Keep only the cases that meet a condition
drop if year == 1996	Drop the cases that meet a condition
keep var1 var2	Keep only these variables (drop var1 to drop one)
sample 20	Keep a random 20% of cases (sample 50, count for exactly 50)
set seed 12345	Run before sample so your random draw comes out the same each time
collapse (mean) var, by(group)	Replace the data with one row per group
preserve … restore	Take a snapshot of the data, do something to it, then get the original back

Conditions use == (equals), != (not equal), >, <, >=, <=, & (and), and | (or). Missing values are stored as . and Stata treats them as larger than any number, so keep if x > 10 also keeps the missings. Add & x < . to leave them out.

Graphs

Command	What it does
histogram var	Histogram of a continuous variable
histogram var, discrete percent	Histogram of a variable with only a few whole-number values
graph bar (count), over(var)	Bar chart of a categorical variable
scatter y x	Scatterplot
scatter y x || lfit y x	Scatterplot with a best-fit line
twoway (scatter y x) (lfit y x)	Scatterplot with a best-fit line (alternate)
graph box y, over(group)	Box plots, one per group
graph export "plot.png", replace	Save the most recent graph to a file

Add titles and axis labels to any graph with options at the end of the command: , title("...") xtitle("...") ytitle("..."). Doing it in the command rather than the graph editor means it ends up in your .do file.

Regression & Tests

Command	What it does
reg y x	Regress y on x (regress)
reg y x1 x2	Multiple regression
predict yhat	After reg: the predicted value for every case
predict resid, residuals	After reg: the residual for every case
ci means var	Confidence interval for a mean (add , level(90) for 90%)
ttest var == 30	One-sample t-test against a specific number
ttest var, by(group)	Difference-in-means t-test (add unequal if the spreads differ)
tab var1 var2, chi2 expected	Chi-square test on a crosstab, showing expected counts
tabi 16 4 \ 10 9 \ 6 18, chi2	Chi-square test on a table you type in (\ starts a new row)

Stata as a Calculator

display (or just di) prints the result of any expression, so you can do hand calculations without leaving Stata.

Command	What it does
di (1820 - 1509) / 321	Arithmetic (here, a z-score)
di normal(0.97)	Area to the left of z = 0.97 under the Normal curve
di 1 - normal(0.97)	Area to the right of z = 0.97
di invnormal(0.975)	The z-score with 97.5% of the area below it (1.96)
di invttail(35, 0.025)	Critical t with df = 35 and 2.5% in the upper tail (for a 95% CI)
di sqrt(36)	Square root
di _b[x]	After reg: the slope on x (_b[_cons] is the intercept)