This guide gives users an introduction to Stata. The topics covered are importing, exploring, modifying and managing data. It has been created using Stata IC version 16.1. The main dataset used is the flights dataset. It contains the US domestic flights in January 2020 . The other two datasets used are fabricated datasets created for the purpose of this guide. For additional support, fill out the support request form.
There are three ways to run tasks in Stata. You can use point-and-click to execute tasks using the dialogs from the Data, Graphics and Statistics menus.
You can type Stata code in the command window and submit the code by hitting the ENTER/return key.
Or you can type Stata code in the do-file editor. To open a do-file editor, go to the File menu, then go to New and select Do-file. Once you type your code in the do-file, you can run it by highlighting the line of code and clicking on the execute icon (the play button) in the top-right corner of the do-file window.
The results will always appear in the results window. Stata is case-sensitive.
Some useful commands to get started:
search enter_topic_here returns the list of commands associated with a topic
help enter_command_here returns the help file for a specific command
clear removes the dataset that is currently loaded in memory so that you can import a new dataset
The working directory is the home directory that Stata works from by default. Pwd gives the current directory. Use cd to change the directory to the location of your data sets. Dir and ls both show the content of the current directory.
pwd cd "[enter path here]" dir ls
The use command can be used to import a Stata dataset. Every time you import a dataset, it is good practice to add the clear option to clear any dataset that is loaded in memory. Options are added after the comma.
use flights, clear
Once you import the dataset, the variables are listed in the variables window. You can check the number of rows and the number of observations in the properties window.
Describe enumerates the list variables, their data type and their labels. Codebook gives a frequency table for character variables, basic statistics (mean, std dev etc) for numeric variables and the number of missing observations for both types of variables.
List can be used to print data values in the results window. It allows users to specify the variables and observations to print.
list originstate in 1 list originstate deptime depdelay in 1000/1005
Tab is an abbreviation for tabulate. The m option is used to display the number of missing values for the specified variables. The cell, row and col options display the cell, row and column percentages.
tab depdelay, m tab depdelay arrdelay, m tab depdelay arrdelay, m cell row col
Sum is an abbreviation for summarize. It gives summary statistics of numeric variables. We can add conditions to a line of code using the if qualifier. For example, we can get the summary statistics of flight distance for flights that took place on Mondays only using the dayofweek variable and setting it equal to 1 which represents Monday. Two equal signs are used for the equal to condition.
sum distance sum distance if dayofweek==1
Bysort can be used to run a line of code for every category of a second variable. For example, we can get the summary statistics of flight distance for each day of the week.
bysort dayofweek: sum distance
The hist, scatter and graph bar commands can be used to make a histogram, a scatterplot and a bar chart respectively. There are many options that can be added to modify all graphs. These options come after the comma. The list of options possible can be found in the help file of each command. Some options can be applied for all or most graphing commands. Title() can be used to add a main title. Xtitle() is used to specify the label of the x-axis. Ytitle() is used to specify the label of the y-axis.
The normal option adds a normal curve on the histogram. Bin() allows the user to specify the number of bins. Color() and lcolor() are used to specify the color of the bin and the border around the bin.
hist deptime hist deptime, normal bin(30) color(ltblue) lcolor(white) xtitle(Departure Time)
The msymbol(), msize() and mcolor() options can be used to specify the shape, size and color of each point on the scatterplot. The xlabel() is specify the ticks on the x-axis. The plotregion() is used to specify the margin between the plot region and the axes.
scatter deptime dayofweek scatter deptime dayofweek, msymbol(+) msize(huge) mcolor(maroon) xlabel(1(1)7) plotregion(margin(10 10 2 2))
The bar chart is usually used to plot summary statistics on the y-axis such as the number of observations for each category of another variable. The count statistic is specified in parenthesis to make a bar chart of the count of observations. The variable on the x-axis is specified in the over() option. B1title() is used to label the x-axis of bar charts instead of xtitle().
graph bar (count), over(dayofweek) graph bar (count), over(dayofweek) title(Frequency of Flights by Day of the Week) ytitle(Frequency) b1title(Day of Week)
You can continue a line of code over multiple lines by adding three forward slashes at the end of the lines that continue.
To generate a new variable that is a function of other variables, use the generate command. Below, the variable distancemiles (distance in units of miles) is created by converting distance in kilometers to miles.
* Example 1 gen distancemiles = distance * 0.621 sum distance distancemiles
* Example 2 gen instate = . replace instate = 1 if originstate==deststate replace instate = 0 if originstate!=deststate tab instate, m
Labels can be created for the entire dataset, for a variable and for the levels of a categorical variable. To label the levels of the variable instate, first the label instatelabel is created using the label define command. Then it is assigned to the variable instate using the label values command.
label define instatelabel 1 "Within State" 0 "Between State" label values instate instatelabel tab instate, m
* Example 3 gen delay = "" replace delay = "not delayed" if depdelay==0 & arrdelay==0 replace delay = "delayed at departure only" if depdelay==1 & arrdelay==0 replace delay = "delayed at arrival only" if depdelay==0 & arrdelay==1 replace delay = "delayed at both" if depdelay==1 & arrdelay==1 tab delay, m
The command egen is used to generate new variables using built-in functions. For example, the mean() function generates mean. Example 4 creates a new variable called meandistance that is the mean of distance. Example 5 also creates a new variable mean meandistancebycarrier which is the mean distance for each carrier or airline.
* Example 4 egen meandistance = mean(distance) tab meandistance
* Example 5 egen meandistancebycarrier = mean(distance), by(carrier) tab meandistancebycarrier
The original flights dataset now has 5 new variables. This can be checked in the variables window. The save command can be used to save this flights dataset. It is good practice to use the replace option.
save flights.dta, replace
Keep can be used to extract a part of the dataset. You can extract specific variables or observations based on criteria.
* Subsetting Data : Variables use flights, clear keep originstate origin deptime depdelay deststate save departure, replace
* Subsetting Data : Observations use flights, clear keep if deststate=="Hawaii" & dayofmonth==1 save hawaii, replace
To merge two datasets, sort each one by the merging variable first then use the merge command.
use flights, clear sort carrier save flights, replace
import excel using "airlinecodes.xlsx", clear firstrow sort carrier merge carrier using flights save flightsmerged, replace
tab _merge tab airline diverted, row
use flights, clear bysort origin : gen departureflights = _N bysort dest : gen arrivalflights = _N gen airport = origin
preserve collapse (first) departureflights (first) arrivalflights, by(airport) export excel "Airports.xlsx", replace firstrow(var) restore
To reshape a data set between wide and long format, use the reshape long or reshape wide commands. For a data set in wide format, the variables that denote the repetition should be named with the variable name followed by a number (i.e. variablename#). For example, in our data set we have drug1, drug2 and drug3. The numbers do not need to be consecutive.
clear use https://mdl.library.utoronto.ca/sites/default/files/statawide list
describe list reshape long drug, i(id) j(drugtype) list
table drugtype, contents(n drug mean drug sd drug)
The following data consists of two dates: one in string format and another one is numeric.
use https://mdl.library.utoronto.ca/sites/default/files/statadates, clear list
To convert string type or numeric dates into elapsed dates or number of days since January 1, 1960 use the date and mdy functions respectively.
generate birthday = date(bdate, "MDY") generate event = mdy(dmon, dday, dyear) drop bdate - dyear list
Note that the third birthday is missing for the variable birthday because the year is not in four digits. This can be fixed manually by replacing the value of birthday for the 3rd observation using the replace command.
replace birthday = mdy(1, 21, 1972) in 3
The format command can be used to control the display of elapsed dates.
format birthday event %dD_m_CY list
The number of days elapsed can be converted into weeks, months and years as follows.
generate diff = event - birthday generate weeks = diff/7 generate months = diff/30.5 generate years = diff/365.25 list
The ssc install command can be used to install packages from the SSC archive.
ado dir ssc install outreg2 ado dir
 The flights dataset is modified from the original version on the Kraggle website.
 Additional resources to learn Stata.
 The tutorial code and the workshop Powerpoint presentation can be found here.
author Nadia Muhe, Map and Data Library, University of Toronto.