0% found this document useful (0 votes)
89 views8 pages

Agenda: Bias Examples Other Examples

1. The document discusses omitted variable bias in multivariate regression analysis. When an important variable is omitted, it can bias the estimated coefficients on the included variables. 2. Several examples are provided to illustrate how omitted variable bias can occur. When a positive correlated omitted variable is left out, it results in upward bias. When a negatively correlated omitted variable is left out, it results in downward bias. 3. The effects of omitted variable bias are demonstrated through examples looking at the relationship between class attendance and grades, education and crime, and graduate education and wages. In each case, omitting a key variable biases the estimated coefficients.

Uploaded by

vaquero92
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
89 views8 pages

Agenda: Bias Examples Other Examples

1. The document discusses omitted variable bias in multivariate regression analysis. When an important variable is omitted, it can bias the estimated coefficients on the included variables. 2. Several examples are provided to illustrate how omitted variable bias can occur. When a positive correlated omitted variable is left out, it results in upward bias. When a negatively correlated omitted variable is left out, it results in downward bias. 3. The effects of omitted variable bias are demonstrated through examples looking at the relationship between class attendance and grades, education and crime, and graduate education and wages. In each case, omitting a key variable biases the estimated coefficients.

Uploaded by

vaquero92
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 8

Lecture 10 - Economics 113

Professor Spearot
I
Agenda
1. Bias Examples
2. Other Examples
I
Exam Reminders
1. No graphing calculators or notes
2. Be on time!!!
Multivariate Regression
Omitted variable bias
I
What happens when we omit an important variable?
I
Need to conjecture regarding the relationship between the omitted
variable and included x and y variables
I
The Table
Corr(omitted variable,x)
positive negative
Corr(omitted variable,y) positive upward bias downward bias
negative downward bias upward bias
I
Upward bias:
I
Estimate is higher than the true parameter: <
b

I
Downward bias:
I
Estimate is lower than the true parameter:
b
<
Multivariate Regression
Omitted variable bias - Examples
I
Example: Eect of class attendance on grades
I
Population follows:
nal =
0
+
1
attend +
2
study + u
I
We instead forget about study and estimate:
d
nal =
b

0
+
b

1
attend
I
Suppose we estimate
b

1
> 0, and conclude that attendance
increases your grade (
1
> 0). Is this right?
I
Positive correlation between study and nal
I
Positive correlation between study and attend
I
b

1
suers from an upward bias.
1
<
b

1
Multivariate Regression
Omitted variable bias - Examples
I
Intuition
I
b

1
> 0 suggests that higher attendance improves your grade
I
However, students who attend class often tend to study more
I
Thus, attend may actually be accounting for the eects of studying,
and not attendance.
I
Overall, given
1
<
b

1
, the result
b

1
> 0 is insucient to guarantee
that
1
> 0.
Multivariate Regression
Omitted variable bias - Examples
I
Example: Eect of drugs on crime
I
Population follows:
crime =
0
+
1
educ +
2
drugs + u
I
We instead forget about drugs and estimate:
[
crime =
b

0
+
b

1
educ
I
Suppose we estimate
b

1
< 0, and conclude education reduces your
likelihood of committing a crime (
1
< 0)
I
Positive correlation between drugs and crime
I
Negative correlation between drugs and educ
I
b

1
suers from an downward bias.
b

1
<
1
Multivariate Regression
Omitted variable bias - Examples
I
Intuition
I
b

1
< 0 suggests that education reduces your likelihood of
committing a crime
I
However, people who go to school are less likely to abuse drugs
I
Thus, educ may actually be accounting for the propensity of drug
use, not the eects of education
I
Overall, given
b

1
<
1
, the result
b

1
< 0 is insucient to guarantee
that
1
< 0.
Multivariate Regression
Omitted variable bias - Examples
I
Example: Eect of graduate education on wages
I
Population follows:
log(wage) =
0
+
1
geduc +
2
Exper + u
I
We instead forget about Exper and estimate:
\
log(wage) =
b

0
+
b

1
log(geduc)
I
Suppose we estimate
b

1
> 0, and conclude that graduate education
increases your wage (
1
> 0)
I
Positive correlation between Exper and log(wage)
I
Negative correlation between Exper and geduc (by construction)
I
b

1
suers from an downward bias.
b

1
<
1
Multivariate Regression
Omitted variable bias - Examples
I
Intuition
I
b

1
> 0 suggests that graduate education of some sort increases your
wage
I
However, people who pursue graduate education have lower levels of
experience
I
Thus, people with no graduate education may earn relatively high
wages since they have lots of experience.
I
Overall, given
b

1
<
1
, the result
b

1
> 0 is sucient to guarantee
that
1
> 0.

You might also like