Scribd
Upload
9 views5 pages

6 ANI Probit and Logit Models Example

The document presents an analysis of factors influencing health insurance purchases using Probit and Logit models based on data from the Health and Retirement Study. Key independent variables include retirement status, age, health status, income, education, marital status, and ethnicity, with results indicating that retired individuals and those in good health are more likely to have insurance, while Hispanics are less likely. The models show similar marginal effects, with an average predicted probability of 38% for having insurance, aligning closely with actual data.

Uploaded by

Miantsaniaina Ramanantsoa
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
9 views5 pages

6 ANI Probit and Logit Models Example

The document presents an analysis of factors influencing health insurance purchases using Probit and Logit models based on data from the Health and Retirement Study. Key independent variables include retirement status, age, health status, income, education, marital status, and ethnicity, with results indicating that retired individuals and those in good health are more likely to have insurance, while Hispanics are less likely. The models show similar marginal effects, with an average predicted probability of 38% for having insurance, aligning closely with actual data.

Uploaded by

Miantsaniaina Ramanantsoa
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/ 5

Probit and Logit Models Example

Ani Katchova

© 2013 by Ani Katchova. All rights reserved.


Probit and Logit Model Example

 We study the factors influencing the purchase of health insurance.


 Using data set from the Health and Retirement Study (HRS), wave 5 (2002) collected by the
National Institute of Aging.
 Dependent variable: whether or not a person has health insurance (0 or 1).
 Independent variables: retired, age, good health status, household income, education years,
married, Hispanic.
 Estimating regression model, logit, and probit models.

Health insurance y codes Percent


frequency
Yes 1 39%
No 0 61%
Binary outcome model coefficients
Have health insurance Regression Logit Probit
coefficients coefficients coefficients
Retired 0.04* 0.19* 0.11*
Age -0.002 -0.01 -0.008
Good health status 0.06* 0.31* 0.19*
HH income 0.0004* 0.002* 0.001*
Education years 0.02* 0.11* 0.07*
Married 0.12* 0.57* 0.36*
Hispanic -0.12* -0.81* -0.46*
Constant 0.12 -1.71* -1.06*
R2 0.08 0.07 0.07
* Indicates significance at the 5% level.

 Coefficient interpretation: Retired individuals (in comparison to non-retired individuals),


individuals with good health status, higher household income, higher education, married are
more likely to have health insurance, and Hispanic are less likely to have health insurance.
 The regression, logit and probit coefficients differ by a scale factor (and therefore we cannot
interpret the magnitude of the coefficients).
Binary outcome model marginal effects
Have health insurance Regression Logit Logit Probit Probit
marginal marginal average marginal average
effects effects at marginal effects marginal
the mean effects at the mean effects
Retired 0.04* 0.04* 0.04* 0.04* 0.04*
Age -0.002 -0.003 -0.003 -0.003 -0.003
Good health status 0.06* 0.07* 0.06* 0.07* 0.06*
HH income 0.0004* 0.0005* 0.0005* 0.0004* 0.0004*
Education years 0.02* 0.02* 0.02* 0.02* 0.02*
Married 0.12* 0.12* 0.12* 0.13* 0.12*
Hispanic -0.12* -0.16* -0.16* -0.16* -0.15*
 Marginal effects interpretation: Retired individuals are 4% more likely to have insurance (in
comparison with those that are not retired). For each additional year in education, individuals
are 2% more likely to have insurance. Hispanics are 16% less likely to have insurance than
non-Hispanics.
 Note that unlike the coefficients which are different, the marginal effects are almost identical in
the three models.
 The marginal effects at the mean and the average marginal effects are almost identical.
 The signs of the coefficients and marginal effects are the same for the logit and probit models.
 The average of predicted probabilities for having insurance is about 38% which is similar to
the actual frequency for having insurance.
 The logit and probit models correctly predict 62% of the values and the rest are misclassified.

You might also like