Stata - Basic Stata Use

Details Overview of Windows within Stata How do I set my working directory? How do I calculate means, variances, and standard deviations? How do I delete observations from a data set? How do I have Stata report normal tail areas and inverse normal tail areas? How do I calculate the correlation between two variables? How do I calculate the correlation between two variables? How do I generate a list of random numbers from a uniform distribution? How do I calculate fitted values and residuals from a regression? How do I divide a variable by a constant (or multiply it or add to it)?

Details  

Overview of Windows within Stata  

When Stata is launched, there are several windows that can be used while working with the program:

  • Command - provides an input area for entering Stata command statements.
  • Results - shows the commands entered along with their results. (If “more” appears at the bottom, press Enter to continue scrolling the results or ‘q’ to quit.)
  • Review - displays a history of the most recent commands entered. Highlighting a command here will make it appear in the Command window.
  • Variables - lists the variables contained in the current data set. Only one data set at a time can be current in Stata. Highlighting a variable in this window will make it appear in the Command window.
  • Data Editor - displays the current data set in spreadsheet format. The data editor can also be used when cutting and pasting data to and from other spreadsheet applications like Excel.
  • Graph - displays the results of graphics associated command statements.

How do I set my working directory?  

Your working directory is the directory or folder in which Stata looks when you give it a disk access command.

  • To see what your current working directory is, type pwd
  • To see the files and folders in your working directory, type ls
  • To move up one level in your directory tree, type cd ..

For instance, if you are in /Home/Users/johndoe/Stata and you type cd .. then you will be in /Home/Users/johndoe.

If your current working directory has a folder called “myfolder” in it and if you want to change your working directory to “myfolder,” type cd myfolder. This, for example, could move you from “/Home/Users/johndoe/Stata” to /Home/Users/johndoe/Stata/myfolder."

Important note! If your folder name has a space in it, you have to enclose the name in quotes. For example: cd "My Folder”

You can use multiple cd .. and cd NAMEHERE commands to move anywhere you want in your hard disk (NAMEHERE refers to a folder into which you want to move; remember to enclose NAMEHERE in quotes if necessary).

How do I calculate means, variances, and standard deviations?  

Use the command summarize. You can simply type summarize, in which case you will get means, standard deviations, and so forth for all variables in memory. Or, you can type

summarize VARNAME

which will give you a summary of the variable VARNAME. Also, add the option detail, as in summarize, detail or

summarize VARNAME, detail

to get various percentiles.

How do I delete observations from a data set?  

Use the drop command. Suppose that a data set has 10 observations. If you type drop in 5 then the 5th observation will be deleted. Similarly, you can type drop in 1/3 to drop the first three observations. Another way to drop delete observations is to use an if clause.

For example, drop if VARNAME<4 will drop all observations that have VARNAME<4. One could have more complication expressions like drop if VARNAME1<5 & VARNAME2>5 and so forth.

How do I have Stata report normal tail areas and inverse normal tail areas?  

To compute the left tail area for a given z value, use the following command:

display normal(z)

where z is the value of interest.

To compute the inverse tail area for an area equal to p, use the following command:

display invnormal(p)

The use of y is generic, and any acceptable label will work. How do I use Stata to calculate tail areas and critical values for the t distribution?

Use the function ttail(n,t) where n is degrees of freedom and t is the critical value of interest. Also, use the function invttail(n,p) where p is a right tail area from a t distribution with n degrees of freedom. For example,display ttail(5,2)will return the upper tail area (to the right of 2) of a t distribution with 5 degrees of freedom. Similarly,display invttail (5,.05) will return the critical value from a t distribution with 5 degrees of freedom such that the area to the right of the value is 0.05. NOTE: Stata is very picky about spaces. Make sure that function calls do not have spaces after arguments. In other words, ttail (5,2) will generate an error but ttail(5,2) will not.

How do I calculate the correlation between two variables?  

Use the command correlate as in

correlate VARNAME1 VARNAME2

which will produce a 2 x 2 correlation matrix. One can also type

correlate VARNAME1 VARNAME2 ... VARNAMEk

which will produce a k X k correlation matrix. Typing correlate without any arguments produces a correlation matrix for all variables.

How do I calculate the correlation between two variables?  

Use the command correlate as in

correlate VARNAME1 VARNAME2

which will produce a 2 x 2 correlation matrix. One can also type

correlate VARNAME1 VARNAME2 ... VARNAMEk

which will produce a k X k correlation matrix. Typing correlate without any arguments produces a correlation matrix for all variables.

How do I generate a list of random numbers from a uniform distribution?  

The command generate x = uniform() will draw random values from the unit interval. These values can be scaled to arbitrary intervals, i.e., “generate x1 = x*10” after assigning x as before will generate x1 from a uniform distribution on the interval from zero through ten. How do I calculate confidence intervals?

Use the command ci. For example, to make a 98% confidence interval for a continuous variable called VARNAME, enter ci VARNAME, level(98). Note the use of the level option to specify the level of the desired confidence interval. If your variable is binary as opposed to continuous, i.e., consists of zeroes and ones, then add the option binomial as in ci VARNAME, binomial level(98). There are various options for how to calculate binomial confidence intervals; these are described in online help.

How do I calculate fitted values and residuals from a regression?  

After a successful regress command, the command

predict r, residual

will create a new variable r that contains residual values. Any variable name may be used. Similarly,

predict yhat

will create a new variable yhat that contains fitted values. As before, any variable name may be used.

How do I divide a variable by a constant (or multiply it or add to it)?  

If the variable is called var1, then the command

replace var1 = var1 / k

changes var1 to var1 / k where k is a specified constant. You could also use

replace var1 = k * var1

or

replace var1 = k + var1

if you wanted to multiply var1, add to it, and so forth. I want to calculate a chi-squared statistic for some tabulated data. I have the table but do not have the data that generated it. How can I have Stata calculate the statistics I need?

Use the tabi command. For example, the command

tabi 9 7 \ 13 7\ 29 26\ 50 47 \ 41 49, chi2

will make a 5 x 2 table using the provided counts and will calculate the chi-squared statistic.

On this page