Linear Regression with Stata

• (brief) Introduction to the Stata Interface
• 2 Examples of Linear Regression
• Exporting Regression Tables
• Where to get help

• tinyurl.com/tischstata

• Open Stata: Windows -> Data & Statistical Applications -> Stata 14

• In Stata: File -> Change Working Directory...

• Command Line
• Review
• Output/Results
• Variables
• Properties

Highly recommended to keep a do/log file of your sessions

• do file = a script of your syntax that can be executed
• log file = record of your output

How Many Variables?

How Many Variables?

Six Variables: region, country, popgrowth, lexp, gnppc, safewater

/* this is a comment, does not run */
use lifeexp.dta

/* basic information about data types */
describe

/* basic descriptive statistics about our dataset */
summarize

browse

Why are values different colors?

Matrix plots

graph matrix lexp safewater gnppc popgrowth

regress y x

regress lexp safewater

regress y x

regress lexp safewater

regress y x

regress lexp safewater

• Each 1 unit increase in safewater estimates a .238 change in life expectancy

scatter plot with fit line

twoway scatter lexp safewater || lfit lexp safewater

clear
use heights.dta
summarize

six variables: earn, height, gender, ed, age, race

Is there a significant relationship between height and salary?

regress earn height

Interpret the output on your own or with a neighbor

Is there a significant relationship between height and salary?

regress earn height

• - each 1 unit increase in height estimates a 1,222 increase in earnings

Multiple Regression

regress earn height ed

what happens to our coefficients when we add variables to the model?

Multiple Regression with a Categorical Variable

regress earn height ed gender

Multiple Regression with a Categorical Variable

/*one way to create a dummy variable*/
gen female = (gender == "female")

/*and another*/
gen female = 0
replace female = 1 if gender == "female"

/*yet another*/
xi: regress earn i.gender

Multiple Regression with a Categorical Variable

regress earn height ed female

outreg2

outreg2 using model.doc
regress earn height 

outreg2

regress earn height ed age
outreg2 using model.doc, append

• Regression Diagnostics
• Hierarchical Models (ANOVA)
• Non-Linear or Non-parametric Models

• In Stata

help regress

• Learning Resources
  • http://researchguides.library.tufts.edu/data/stata

email: joshua.quan@tufts.edu