Scribd
Upload
8 views

Stata Workshop

1. This document provides an overview of using Stata for statistical analysis and includes descriptions of key Stata windows, file types, commands, and exercises. 2. It describes how to open and save Stata data files, create new variables, estimate regressions, and generate statistics and graphs with both the interactive menus and command lines. 3. The exercises provided guide users through loading sample data, generating wage and other variables, computing summary statistics both with and without weights, estimating a regression of wages on education, and checking properties of the regression results.

Uploaded by

hasan zahid
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
8 views

Stata Workshop

1. This document provides an overview of using Stata for statistical analysis and includes descriptions of key Stata windows, file types, commands, and exercises. 2. It describes how to open and save Stata data files, create new variables, estimate regressions, and generate statistics and graphs with both the interactive menus and command lines. 3. The exercises provided guide users through loading sample data, generating wage and other variables, computing summary statistics both with and without weights, estimating a regression of wages on education, and checking properties of the regression results.

Uploaded by

hasan zahid
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

ECO311, Spring 2019

Prof. Bill Even

STATA Workshop

In addition to the online help menu in Stata, there are numerous resources on the web to help you
learn stata. I've posted links to a few good sites under the module titled “Stata Resources” in
Canvas.

The stata windows.


 command (for interactive submission of commands)
 results
 variables
 review

Stata icons
 log (to create a file containing a log of session)
 do-editor (to create a "*.do" file containing list of commands to be executed)
 data editor (to view and/or edit stata data as spreadsheet)
 data browser (to browse, but not edit, stata data files)
 break (to stop execution of Stata commands)
 Stata viewer (to view log files, etc.)

Types of stata files.


 dta (Stata data sets)
 do (Stata files containing commands)
 smcl/log (log files containing Stata commands and results from execution)

1. Different ways to create a Stata data set


a. Use data editor and manually enter
b. Use data editor and cut/paste from excel
c. Import using file/import/…
d. set obs=100 to create empty data set with 100 observations and then create own
variables.

2. Opening an existing Stata data set


a. Interactive: file/open/…
b. Command line: use "G:\ECO\evenwe\eco311\data\cps2016.dta", clear

After importing the data, creating new variables and/or dropping observations, you may want to
save the new data set as a Stata data set. You can save a Stata set interactively with “file/save
as/...”. Alternatively, you could add code in your program so that the Stata data set is saved
every time you run the program. The following would create a stata data set name
“mycps2016.dta”.

3. Saving a Stata data set.


a. Interactive: file/save/….
ECO311, Spring 2019
Prof. Bill Even

b. Command line: save "m:\mycps2016.dta", replace

[Note: you will probably want to create a folder on your m-drive to hold your eco311
files.]

4. The command line versus do-files

You should practice using do-files to accumulate commands for your data creation and
analysis. This makes it possible to accumulate all of your commands and make necessary
changes as you work through your project. You can create the relevant commands using the
drop down menus or typing them interactively, but you should save the commands necessary
for your analysis in a do file. You should save your results in a log file.

5. Creating and replacing variables.

Many mathematical functions are available. Type “help functions” on Stata command line
for assistance.

To generate a new variable, use the gen command. Note: Stata is case sensitive. All
commands are in lower case. I recommend making all your variables in lower case as well.
If you wish to update a variable that has already been created, use the replace command.

Examples:
gen school2=school^2 /*create the square of school years*/

gen age50_=1 if age>=50 /*create a dummy variable for workers over age 50*/
replace age50_=0 if age<50

gen age50_=(age>=50) /*creates a dummy equal to one whenever the condition in


parentheses is true*/
drop if age>50 /*warning: missing values are treated as infinite*/
drop if age==. /*drop observations with missing values for age*/
keep if age<=50

(Note: for comparison operators, must use double equal sign for equality: e.g.

drop if age==50 /*(NOT drop if age=50) */

drop if age==. /*drops all observations with “missing values” for age*/

drop if age~=44 /*drops all observations with age not equal to 44*/
ECO311, Spring 2019
Prof. Bill Even

(other logical operators >= and <=)

drop if age<=16 | age>=65 / *the “or” operator is | */


drop if age>=55 & female==1 /*the “and” operator is “&” */

Note: Stata treats a missing value (“.”) has being infinitely large. Hence, if the
variable age has a missing value, issuing the command “drop if age>50” will drop
all people over age 50 as well as those with missing values for age

6. Viewing the data


a. list
b. list in 1/10
c. list if age>50

7. Create sample statistics


a. summarize
b. tabstat
c. corr
d. table

8. Create statistics by groups


a. bysort female: summarize age
b. table female, c(m age p50 age)

9. Weights
a. Many Stata commands can be adjusted for weights
i. pw=probability weights
ii. aw=analytical weights
iii. fw=final weights

b. examples
i. summarize age [pw=finalwt]

10. Create graphs


a. interactive
b. create interactive and copy command to do-file for later use.

Exercise 1.
a. Open the Stata data set at g:\eco\evenwe\eco311\ cps2016.dta
b. Create a histogram of age
c. Create hrwage (wkearn/wkhours)
d. Compute the mean and median of age and hrwage
a. Without weights
ECO311, Spring 2019
Prof. Bill Even

b. With analytical weights [aw=earnwt]


e. Compute the number of workers in the U.S. [tabstat command, create a dummy for
workers based on _esr and sum the dummy using aweights]
f. Create a variable representing the log(hrwage)
g. Compute the mean of hrwage for workers under age 25
h. Compute the mean of hrwage for workers between age 25 and 35

5. OLS REGRESSIONS.

To estimate an OLS regression of hourly earnings on years of schooling, we use the


procedure called regress.

regress hrwage school

An intercept is automatically included (though it can be dropped as an option).


The regression could be restricted to people aged 55 or less with:

regress hrwage school if age<=55

or the regression could be done separately by sex with:

bysort female: regress hrwage _school

To generate a variable called yhat with predictions from the regression, and uhat with the
predicted residuals:

predict yhat, xb

predict uhat, residual

Predict can also be used to generate residuals, standard errors of the prediction, etc. See help
for regress.

Exericise 2.
a. Estimate a simple linear regression of the hourly wage on years of education.
b. Compute predicted values of hourly wage
c. Compute predicted residuals
d. Show that the sum of the residuals equals zero (use tabstat command)
e. Show that the covariance between school and the residuals is zero (corr command)

f. Show that the regression line “passes through the mean”


a. Note: you can do calculations in Stata by creating and displaying a scalar.
For example,

scalar x=13+14
ECO311, Spring 2019
Prof. Bill Even

display x

will create and display the value of (13+14)

You can do calculations using coefficient estimates as follows:

scalar z=_b[_cons] + _b[age]*40


display z

_b[xxxx] refers to the coefficient from the most recent regression on the
variable xxxx. _cons refers to the interecept in the regression.

You might also like