BIVARIATE REGRESSION
Steven Livingston
Fall 2014
[OLS (Ordinary Least Squares) and Linear Regression are basically synonyms.]
Regression allows us to model, explain, or predict how the world works. Though
you can observe alot by watching as Yogi Berra onc
T-Tests vs. F-Tests
Livingston
PS 6500
T- Tests
1. Why do we do a t-test? To see if the true coefcient of a variable might really
be 0. If it is zero, that variable does not belong in the regression equation. This is
not simple to decide because a coefcie
Multiple Regression
Steven Livingston
Fall 2015
For the most part, multiple regression is a straight-forward extension of bivariate
regression. It simply adds additional independent variables to the model. Were going
to point out the few differences, thou
STATA
a= is the constant in the equation, it explains y when x=0
b=is the slope, it explains how much x influences y, it is the coefficient that expresses the relationship
between x and y
s.e. = the residual indicates the difference between the y and the
*program to monte carlo OLS results
program simple_monte, rclass
gen dis=rnormal(0,1)
gen y = 2 + 5*x + dis creo io l'equazione cosi conosco gia la risposta
mettendo un errore random
regress y x
return scalar b=_b[x]
return scalar seb=_se[x]
retu
Multiple Regression
Steven Livingston
Fall 2014
For the most part, multiple regression is a straight-forward extension of bivariate
regression. It simply adds additional independent variables to the model. Were going
to point out the few differences. So.
Functional Form: Nonlinearity
Steven Livingston
Fall 2014
In OLS regression, Y is assumed to be a linear function of the independent variables.
In the real world, that assumption is often not met. Here is a scatterplot of gdp per
capita against a nations
Key Properties of Regression
Steven Livingston
Fall 2014
We need to look at a few properties of regression now. Mostly for future reference!
The constant in a regression equation is almost always of little interest. Its the slope
coefcient that really mat
Model Assessment & Hypothesis Testing
Steven Livingston
Fall 2014
More important than the generation of a model is the assessment of whether its any
good! This actually involves two things, (1) judging the overall accuracy of the model
and (2) determining
STATA
preserve
restore
drop if year<1900 drop if year=1900 drop y* keep if year=2010
list [variable]
desc [variable]
sum [variable]
tab [variable]
gen [new variable name]=[existing variable]+/- sthg OR ln(existing variable)
F-test
test [variable]
Multicol