Deck 12: Limited Dependent Variables

Question 1

Estimating models with binary dependent variables by OLS is referred to as estimating a

A)logit model.
B)probit model.
C)multinomial logit model.
D)linear probability model.

Answer 1

D

Question 2

Suppose you estimate a logit model explaining the probability of entering the labor force and you get the following estimated coefficients $Lab\widehat {or ~F}orce _ { i } = 4.83 + 1.27 \text { Education } _ { i } - 3.06 \text { Female } _ { i } + 2.91 \text { Children Under 6} _ { i }$
You should conclude that

A)men are estimated to be 306 percent more likely than women to enter the labor force.
B)men are estimated to be more likely than women to enter the labor force.
C)each additional year of education increase the likelihood of entering the labor force by 483 percent.
D)men are estimated to be 3.06 percent more likely than women to enter the labor force.

Answer 2

men are estimated to be more likely than women to enter the labor force.

Question 3

A binary dependent variable is one that takes on

A)continuous values.
B)only the values 0 and 1.
C)only integer values.
D)only non-negative values.

Answer 3

B

Question 4

Suppose you are interested in estimating the effect that getting a flu shot has on the probability of contracting the flu.In this case,your preferred estimator would be

A)OLS.
B)WLS.
C)multiple linear regression.
D)a probit model.

Question 5

The coefficient estimates from the multinomial logit model

A)are constrained to be positive.
B)must fall between 0 and 1.
C)indicate the estimated marginal effects that the independent variables have on the dependent variable.
D)indicate the general degree to which the independent variables are correlated with the dependent variable.

Question 6

Suppose you estimate a logit model explaining the probability of entering the labor force and you convert the estimated coefficients into the following estimated marginal effects $Lab\widehat {or ~F}orce _ { i } = 1.67 + 0.42 \text { Education } _ { i } - 3.06 \text { Female } _ { i } + 2.91 \text { Children Under 6} _ { i }$
You should conclude that

A)men are estimated to be 306 percent more likely than women to enter the labor force.
B)men are estimated to be more likely than women to enter the labor force.
C)each additional year of education increase the likelihood of entering the labor force by 483 percent.
D)men are estimated to be 3.06 percent more likely than women to enter the labor force.

Question 7

OLS is not the preferred model with binary dependent variables because it

A)does not account for negative values.
B)only accounts for positive values.
C)estimates constant marginal effects.
D)does not account for values greater than 1.

Question 8

An estimator that is preferable to OLS in the case of binary dependent variables would be one that constrains all values of the dependent variable to

A)be negative.
B)fall between 0 and 1.
C)equal 0.
D)equal 1.

Question 9

When using the logit model,estimated marginal effects are obtained by

A)estimating the model by the logit function.
B)estimating the model by OLS.
C)taking the derivative of the estimated coefficients.
D)converting the estimated coefficients into marginal effects by taking the derivative of the logit function with respect to the indepdendent variable.

Question 10

The coefficient estimates from the probit model are

A)true marginal effects.
B)non-negative.
C)not true marginal effects.
D)biased.

Question 11

Suppose you estimate a logit model explaining the probability of entering the labor force and you get the following estimated coefficients $Lab\widehat {or ~F}orce _ { i } = 4.83 + 1.27 \text { Education } _ { i } - 3.06 \text { Female } _ { i } + 2.91 \text { Children Under 6} _ { i }$
You should conclude that

A)men are estimated to be 306 percent more likely than women to enter the labor force.
B)that a higher education level increases the probability of being in the labor force.
C)each additional year of education increase the likelihood of entering the labor force by 483 percent.
D)men are estimated to be 3.06 percent more likely than women to enter the labor force.

Question 12

The coefficient estimates from the logit model are

A)not true marginal effects.
B)true marginal effects.
C)non-negative.
D)biased.

Question 13

The coefficient estimates from the probit model

A)indicate the estimated marginal effects that the independent variables have on the dependent variable.
B)indicate the general degree to which the independent variables are correlated with the dependent variable.
C)are constrained to be positive.
D)must fall between 0 and 1.

Question 14

When using the probit model,estimated marginal effects are obtained by

A)estimating the model by the probit function.
B)converting the estimated coefficients into marginal effects by taking the derivatives of the probit function with respect to the independent variable.
C)estimating the model by OLS.
D)taking the derivative of the estimated coefficients.

Question 15

The multinomial logit is preferable to OLS in the case of categorical dependent variables because it

A)is constrained to be non-negative.
B)accounts for the categorical nature of the independent variables.
C)accounts for the categorical nature of the dependent variable.
D)unbiased.

Question 16

Categorical dependent variables are ones that take on

A)continuous values.
B)only the values 0 and 1.
C)only a few integer values.
D)only non-negative values.

Question 17

Suppose you estimate a logit model explaining the probability of entering the labor force and you get the following estimated coefficients $Lab\widehat {or ~F}orce _ { i } = 4.83 + 1.27 \text { Education } _ { i } - 3.06 \text { Female } _ { i } + 2.91 \text { Children Under 6} _ { i }$
You should conclude that

A)men are estimated to be 306 percent more likely than women to enter the labor force.
B)men are estimated to be more likely than women to enter the labor force.
C)each additional year of education increase the likelihood of entering the labor force by 483 percent.
D)men are estimated to be 3.06 percent more likely than women to enter the labor force.

Question 18

Suppose you are interested in estimating the effect that getting a flu shot has on the probability of contracting the flu.In this case,your preferred estimator would be

A)OLS.
B)a logit model.
C)WLS.
D)multiple linear regression.

Question 19

The coefficient estimates from the logit model

A)indicate the estimated marginal effects that the independent variables have on the dependent variable.
B)indicate the general degree to which the independent variables are correlated with the dependent variable.
C)are constrained to be positive.
D)must fall between 0 and 1.

Question 20

The coefficient estimates from the multinomial logit model are

A)non-negative.
B)not true marginal effects.
C)true marginal effects.
D)biased.

Question 21

Suppose you are interested in explaining Student Evaluations of Teaching (which are calculated on the following 4-point scale: (1)Poor, (2)Fair, (3)Good,and (4)Excellent)in a sample of 15,239 college students and that using the multinomial logit you obtain the following coefficient estimates (standard errors in parentheses)
a)How should you interpret the estimated coefficient on Average GPA? Does this make economic sense? Explain.
b)How should you interpret the estimated coefficient on % Freshmen? Does this make economic sense? Explain.
c)Do these estimates provide all of the information that you desire? If not,how can you obtain that information? Explain.

Question 22

What do the coefficient estimates from the multinomial logit model represent? What do you need to do to determine true estimated marginal effects? Explain.

Question 23

Suppose you are interested in estimating the effect that gasoline prices have on the decision to ride a bike,take a bus,take a train,or drive your own car to work in this case,your preferred estimator would be

A)OLS.
B)a logit model.
C)a multinomial logit model.
D)a probit model.

Question 24

When using the probit model,estimated marginal effects are obtained by

A)converting the estimated coefficients into marginal effects by plugging the values back into the multinomial logit function.
B)estimating the model by OLS.
C)estimating the model by the probit function.
D)taking the derivative of the estimated coefficients.

Question 25

What is a logit model? Why is it more appropriate than OLS? Explain.

Question 26

What is a multinomial logit model? Why is it more appropriate than OLS? Explain.

Question 27

The ordered probit is more appropriate than the multinomial logit

A)when the dependent variable only taken on two values.
B)when the independence of irrelevant alternatives is violated.
C)always when dealing with categorical dependent variables.
D)always when dealing with a non-multinomial dependent variable.

Question 28

What do the coefficient estimates from the logit model represent? What do you need to do to determine true estimated marginal effects? Explain.

Question 29

What do the coefficient estimates from the ordered probit model represent? What do you need to do to determine true estimated marginal effects? Explain.

Question 30

What is a categorical dependent variable? Why does it present a challenge to estimating the population multiple linear regression model by OLS? Explain.

Question 31

Suppose you are interested in explaining Student Evaluations of Teaching (which are calculated on the following 4-point scale: (0)Poor, (1)Fair, (2)Good,and (3)Excellent)as a function of Average GPA,% Freshmen,% Male,and % Majors.
a)Write out the regression model that you would want to estimate.Explain the dependent variable in detail.
b)Could you estimate the above model by OLS? Are there any potential shortcomings of estimating such a model? If so,what are they? Explain.
c)In reference to your answer in (b),is there an alternative estimator that is more preferred? If so,what is it? Why is it preferred to the model in (b)? Explain.

Question 32

Suppose you are interested in explaining the probability of Switching Jobs as a function of Salary,Experience,Home Ownership,and Number of Children.
a)Write out the regression model that you would want to estimate.Explain the dependent variable in detail.
b)Could you estimate the above model by OLS? What would you call such a model? Are there any potential shortcomings of estimating such a model? If so,what are they? Explain.
c)In reference to your answer in (b),is there an alternative estimator that is more preferred? If so,what is it? Why is it preferred to the model in (b)? Explain.

Question 33

What is the ordered probit model? Why is it more appropriate than the multinomial logit? Explain.

Question 34

Suppose you are interested in explaining the probability of Switching Jobs in a sample of 32,517 men and women and that using the logit you obtain the following coefficient estimates (standard errors in parentheses)
a)How should you interpret the estimated coefficient on Salary? Does this make economic sense? Explain.
b)How should you interpret the estimated coefficient on Rent Home? Does this make economic sense? Explain.
c)Do these estimates provide all of the information that you desire? If not,how can you obtain that information? Explain.

Question 35

Suppose you are interested in explaining the probability of Switching Jobs in a sample of 32,517 men and women and that using the probit you obtain the following coefficient estimates (standard errors in parentheses)
a)How should you interpret the estimated coefficient on Salary? Does this make economic sense? Explain.
b)How should you interpret the estimated coefficient on Rent Home? Does this make economic sense? Explain.
c)Do these estimates provide all of the information that you desire? If not,how can you obtain that information? Explain.

Question 36

The assumption required for the multinomial logit to be the most-preferred estimator for categorical dependent variables is referred to as

A)multinomial logit theorem.
B)independence of relevant alternatives.
C)independence of irrelevant alternatives.
D)dependence of irrelevant alternatives.

Question 37

What is a binary dependent variable? Why does it present a challenge to estimating the population multiple linear regression model by OLS? Explain.

Question 38

What do the coefficient estimates from the probit model represent? What do you need to do to determine true estimated marginal effects? Explain.

Question 39

What is the independent of irrelevant alternatives? Given an example.Why is it important? Explain.

Question 40

What is a probit model? Why is it more appropriate than OLS? Explain.