Regression with Qualitative

Dependent Variable
Recommended Text: Damodar Gujarati-Econometrics by Example
Chapter 8
DATA
Yi   0  1 X i   i
Where Yi=1 if returned
=0 if not returned
Xi= Age
Post estimation of β0 and β1 , OLS regression gives you the average
value of Yi, given the values of Xi, i.e.

E (Yi | X i )  b0  b1 X i
How to interpret the E (Yi | X i ) when Y is a binary variable?

Yi Probability
1 pi
0 1-pi E (Yi | X i )  1 pi  0(1  pi )  pi
Total 1

Therefore, the expected value of Yi, given the value of Xi, is the
conditional probability that the event will occur (success probability)
given the value of Xi
E (Yi | X i )  P(Yi  1 | X i )  pi
 DATA
Yi  1.296  0.038X i
Regression Statistics
Multiple R 0.735183781
R Square 0.540495192 Age of Club Member Line Fit Plot
Adjusted R Square 0.535389583
1.5
Standard Error 0.334483828
Observations 92
1
ANOVA

Returned
df SS MS F Significance F 0.5 Returned
Regression 1 11.84389465 11.84389 105.863 7.18361E-17 Predicted Returned
Residual 90 10.06914883 0.111879 0
Total 91 21.91304348 0 20 40 60 80

-0.5
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Age of Club Member
Intercept -1.2960182 0.167659972 -7.73004 1.46E-11 -1.629104001 -0.9629324
Age of Club Member 0.03753232 0.003647819 10.28898 7.18E-17 0.030285291 0.044779349
Interpretation

Yi  1.296  0.038X i
For example, If X=53, the probability that the form will be returned
(P(Y=1)) is 0.693.
If X increases by 1 unit (year in this case), the probability that the form
will be returned increases by .038
This means that if Age increases by 1 year, probability of returning the
form will rise from 69.3% to 73.1%, by 3.8 percentage points.
Major Problems with LPM
• Assumes linear relation between probability and values of
explanatory variable.

• Estimated probability value may not always lie between 0 and 1

DATA

return
return

LPM Logit

Y=b0+b1X
E(Y|X)=p  0  1 X
e
f (Y )  p   0  1 X
1 e
 THE LOGIT MODEL

Unlike LPM, the probability of success in

the logit model is described as:

 0  1 X
e
f (Y )  p 
return

 0  1 X
1 e

p is the probability that Y=1

Odds ratio p
S  oddsratio 
1 p
 0  1 X
e
 0  1 X
S 1  e
 0  1 X
e
1
1  e  0  1 X

 0  1 X
S e LOGIT MODEL

ln(S )   0  1 X
Interpreting the Logit Regression result
• The log of the odds ratio or logit equation:

ln(S )  20.4078  0.4259Age

• To find out the probability that a 50 year old person
returns the form:
ln(S )  0.8879

The antilog of this value results in the odds, S:

𝑝
S e 0.8879
 2.43 𝑆=
1−𝑝
= 2.43

𝑝 = .7084
Interpreting the Logit Regression result

Alternatively, the probability value can also be found out using:

20.4078 0.4259 Age

e
p 20.40780.4259 Age
1 e
for Age=50

p  .7084
The marginal effect: change in probability of success due to change in X

ln 𝑆 = 𝑏0 + 𝑏1 𝑋
𝑝
ln = 𝑏0 + 𝑏1 𝑋
1−𝑝
ln p − ln(1 − p) = 𝑏0 + 𝑏1 𝑋
1 𝑑𝑝 1 𝑑𝑝
−( )(−1) = 𝑏1
𝑝 𝑑𝑋 1−𝑝 𝑑𝑋
1 𝑑𝑝 1 𝑑𝑝
−( )(−1) = 𝑏1
𝑝 𝑑𝑋 1−𝑝 𝑑𝑋
𝑑𝑝 1 1
( + ) = 𝑏1
𝑑𝑋 𝑝 1 − 𝑝
𝑑𝑝 1
= 𝑏1
𝑑𝑋 𝑝 1 − 𝑝
𝑑𝑝
= 𝑏1 𝑝 1 − 𝑝
𝑑𝑋
The marginal effect: change in probability of success due to change in X

ln(S )  20.4078  0.4259Age

Age =50
p=.7084
What is the impact on probability of returning the form if age increases
by 1 year?
𝑑𝑝
= 𝑏1 𝑝 1 − 𝑝
𝑑𝑋
𝑑𝑝
= 0.4259 0.7084 1 − 0.7084 =.0879
𝑑𝑋
 CHARACTERISTICS OF LOGIT MODEL
 The interpretation of the logit model is as follows: Each slope coefficient shows
how the log of the odds in favor of the outcome changes as the value of the X
variable changes by a unit.

 Once the coefficients of the logit model are estimated, we can compute the success
probabilities.

 In the logit model, the marginal effect of a unit change in the explanatory variable
not only depends on the coefficient of that variable but also on the level of
probability from which the change is measured. The latter depends on the values
of all the explanatory variables in the model.