StatLab Stata Workshop

Stata Workshop

Spring 2003
Alexandra Guisinger

Using STATA at the Statlab

Click on Stata 7 icon (found under the Start Menu)

The Different Windows in STATA

Automatically displayed windows
- Command Window: executes STATA commands; type in commands here and execute with the ENTER key.
- Results Window: displays commands entered and corresponding output/results; for screen display only; contents cannot be edited.
- Review Window: displays commands already entered; to re-run commands without typing them again, "click" the command line; the highlighted command line will appear in the STATA Command Window; ENTER to execute the command; contents may also be saved to a file to edit and to use as a do-file (click on review menu -- the box to the left of the word review -- and select "Save Review Contents"
- Variable Window: displays the variable names and labels and allows for "point and click" additions to the command line
Other Windows accessible through the "Windows" menu or by clicking on the respective menu button
- Data Editor: allows for manual entry of data or manual correction of data (use cautiously; however you will be asked to confirm changes upon closing the window)
- Data Browser: allows for viewing but no editing of data (button accessible only, safer than Data Editor)
- Do-File Editor: allows for the creation of saved lists of command "do-files"
- Viewer Window: allows for you to open log files which are text file versions of the results window suitable for editing (also provides access to help files)

2. Entering Data

Opening an existing Stata Dataset
- Select Open under the File Menu and browse for the Stata files (.dta)
- Type use "directorypath\filename" [for example: use "i:\workshop\stata\auto.dta"] in the command line . To access files stored on your personal account, the command and address would be something like: use "y:\mydata\newdataset.dta"

Opening another program's dataset or an ASCII dataset

Use StatTransfer (found under the Start Menu) to transform the file into a Stata dataset (.dta). A help file is available on the StatLab webpage (http://statlab.stat.yale.edu).
Alternatively for an ASCII file use the command "infile" and follow directions in the Stata Manuals.

Entering data manually (for example see below)

Open the data editor as described above and either type in data or copy and paste data from an excel spreadsheet (caution: do not enter "." for empty cells; Stata will generate these markers by itself.)
STATA 7 variable names can be up to 32 characters. However, be wary with long names. Stata 6 only recognizes up to 8 characters so long names will make files more difficult to transfer. Also, Stata will attempt to guess the variable when abbrieviates forms of the name are used. Lengthy names may add to confusion or use of a similar variable.

Please note that Stata is not backward compatible, in other words Stata 6 cannot open Stata 7 datafiles. Statlab computers uses Stata 7. If using an older version elsewhere, you should remember to use StatTransfer to convert the datafile.

3. Analyzing Data

Before you do any analysis OPEN A LOG FILE
- You must direct output to a log file to save and/or print the results of your analysis. To do so either:
  - Select Log>Begin from the File Menu and select to save as either log (text) or formatted log (sncl)
  - Type in the command line: log using "c:\temp\mylogname", [append or replace] [text or sncl]
- To close your log either:
  - Select Log>Close
  - Type in the command line: log close
- Logs capture all the text printed in the results window. To save time editing, consider suspending (log off) the log when it is not necessary (log on to resume). Logs can be edited in wordpad etc. or in the viewer.

The Basic Syntax of a STATA command is as follows
- COMMAND Variable-list Restrictions, options
- For Example: regress y x1 x2 x3 if x4==2, noconstant
- In the manual, the generic command lines will look as the following:
  - regress depvar [varlist] [weight] [if exp] [in range] [, level(#) beta robust cluster(varname) hc2 hc3 hascons noconstant tsscons noheader eform(string) depname(varname) mse1 plus ]
  - Text within brackets[] are optional restrictions or options.
  - Underlined sections (bolded above) of the word highlight acceptable abbreviations (i.e. reg instead of regress; noc instead of noconstant)
- Note: Stata is case sensitive
- To add variables to the command line, you may either type the variable or click on the variable name in the variable window
Obtaining information about the data file (for this example using data from i:\ workshop\stata\auto.dta)
- describe to display variable names, types and labels (also whether string or number)
- list to display all the observations and variables in the data file
  Note: Interrupt the --more-- display with q (for quit) or the Break button on the Toolbar (1^st on right) or <CTRL+<BREAK
- Modify list command to display only a subset of the dataset:

list price mpg to display values of price and mpg for all observations
list price mpg in 10/20 to display values of price and mpg for observations 10 through 20
list price mpg if mpg<20 to display values of price and mpg for observations where mpg<20

Descriptive statistics:
- summarize price mpg returns number of observations, mean, standard deviation, minimum and maximum values of the two variables
TIP: to recall previous commands, use the <PageUp and <PageDown keys, or highlight them in the Review Window and ENTER to execute.

Generate/Modify Variables
- generate wgt_sq = weight^2
- generate lowrep = rep78<=3 Creates a new variable lowrep, an indicator variable (0,1) for low repairs.
- Note the difference with generate lowrep = 1 if rep78<=3 . the latter generates missing values"." for all observations with rep78>3
Note: The generate command only works with non-existing variable names.
The replace command modifies existing variables only.

When creating a variable based on a function (min,max,diff, etc) you will need to use the command egen. See help for details.

Graphs

Creating Graphs
- graph rep78 for a histogram
- graph mpg weight, title("Miles per Gallon / Weight") for a scatter plot
- graph mpg weight, yscale(0,.) xscale(0,.) to set scales the axes to start at zero.
Saving Graphs
- graph mpg weight, saving(c:\temp\graph1) saves the graph as c:\temp\graph1.gph
  OR
- graph mpg weight
  choose File/Save Graph from the pull-down menu, navigate to c:\temp and save as graph1.gph
- graph using graph1.gph
  displays the graph saved in file graph1.gph
Printing Graph
- Click on File/Print Graph from the pull-down menu

4. Using Do-Files

Lists of commands can be stored in the form of do-files. These files are useful both as records of data manipulation and of the estimation processes as well as replication on other datasets.
Creating Do-Files
- The review window stores in order all commands entered in the command line. This window can be saved and then edited in the Do-FIle Window.
- Alternatively, you can enter text directly into the do file window. For example:

set more off
log using "c:\temp\Cars_oct26", replace t
use "c:\temp\cardata.dta",clear

describe mpg foreign rep78

summarize mpg foreign rep78

regress mpg foreign rep78

log close

some people prefer to create a line ending marker. To do so type #delimit ; to make a semicolon an end of line marker
To run the do file select either:
- to run with output showing in the results screen
- to run "quietly" i.e. without output showing
To run only a part of the do-file, use cursor to highlight section to be run (if have used delimit statement this must always include this line)

5. Other Helpful commands

set memory : use to increase memory capacity. set memory 100m
set matsize : increases the number of variable allowable from 40 up to 800. set matsize 200
clear: clears existing data in Stata
quietly: suppresses output. qby
replace: overwrites existing files save "c:\temp\myfile,replace"
append: appends to log file
reshape: changes "long" datasets to "wide" datasets and vice versa

6. Exiting STATA

exit OR, if you haven't saved your work, but want to exit anyway: exit,clear

7. Getting Help

STATA Manuals provide detailed descriptions of commands and options along with examples and methods issues ordered by command. Search index to find commands.
On-line help (see help menu) allows for search both by topic and by command, but on-line explications lack examples
For FAQs, Help, and this workshop, see the StatLab software help page http://statlab.stat.yale.edu/statlab/software.html

EXAMPLE: Creating STATA datasets

Restart STATA. If STATA is already loaded, clear data in memory using clear

Manual input
You can input data manually using the Editor. Click on the Editor button on the Toolbar (4^th from the right).

Enter the following 3 columns (make, price and mpg) and 10 lines of data:

AMC Concord      4099        22

AMC Pacer           4749        17

AMC Spirit                 22

Buick Century    4816        20

Buick Electra    7827        15

Note: A missing value (empty data cell) is coded with a dot: "."; you do not need to type in the dot; if you leave the cell blank, it is automatically inserted.

After you have entered the data, double-click anywhere in the column to invoke the STATA variable definition dialogue box
Note: you will not be able to invoke this box until you have entered some data

Enter variable names for your columns: make (instead of var1), price and mpg
Note: variable names can be 32 characters long; also, remember that STATA is case sensitive!

Saving your data in STATA format

save c:\temp\myfile saves data in workspace into a new STATA dataset myfile.dta.

Yale University Social Science Statistical Laboratory
Comments: stathelp@yale.edu
URL: http://statlab.stat.yale.edu