Lately, I have had to read several papers in economics (a field that I'm not too familiar with). One thing that I've noticed is that even when the response variable is binary, linear regression models fitted using OLS are ubiquitous. My question is therefore:
Why is linear regression favoured over for instance logistic regression in the field of economics? Is this simply common practice or is it a procedure that is actively advocated (in papers, by teachers, etc.)?
Please note that I am not asking why using linear regression with a binary response may be a bad idea, or what the alternative methods are. On the contrary, I am asking why people use linear regression in this setting because I know the answers to these two questions.
Best Answer
This blog post by on Dave Giles' econometrics blog mostly outlines the disadvantages of the Linear Probability Model (LPM).
However, he does include a short list of reasons why researchers choose to use it:
- It's computationally simpler.
- It's easier to interpret the "marginal effects".
- It avoids the risk of mis-specification of the "link function".
- There are complications with Logit or Probit if you have endogenous dummy regressors.
- The estimated marginal effects from the LPM, Logit and Probit models are usually very similar, especially if you have a large sample size.
I don't know that the LPM is all that commonly used compared with logit or probit but some of those reasons above are sensible to me.
Similar Posts:
- Solved – theoretical basis for logistic regression
- Solved – How to choose between logit, probit or linear probability model
- Solved – Conditional logistic regression vs GLMM in R
- Solved – use AIC value for comparing logit and probit model where for each model the number of covariates are equal
- Solved – Running a Probit on Survival-Time Data