Sample selection model for continous dependent variable.

[Irfan Ali]

Dear Statalisters,

A few questions have got me confused and I am sure most of you know the answers.

1. I want to test the regression model for sample selection bais but I have a cont. dependent variable (e.g ROA). What i have understood so far that Heckman model is only applicable if the model is tested using probit? Is my understanding of the Heckman correct?

2. Which method i must try for sample selection bais considering i have only continous variable and i cannot run probit, ? The alternatives.

3. Can Heckman model be applied to different(all) models?

Thanks and have a nice day.

Irfan.


p.s. I have a panel dataset.

[Phil Bromiley]

You didn't get a quick answer. You'll increase your chances of a helpful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

The normal Heckman model assumes a binary selection and a continuous dv on the main equation. Stata includes a probit Heckman estimator. If need be, you can do selection models with many variants in the extended regression procedures in recent versions of Stata.

[Irfan Ali]

Thank you Phil.

I tried this code and the womendummy my IV gets ommitted because of collinearity. I am unable to figure it out whats wrong with this code. One thing is clear i shouldn't use women dummy twice but then i dont know how to run it without it?

 heckman MktLEVERAGE WOMENDUMMY  PROFITABILITY GROWTH LIQUIDITY SIZE TANGIBILITY NDTAX MARKET_BOOK , select( WOMENDUMMY = EDUCATION LIFE_EXPECTANCY PROFITABILITY GROWTH LIQUIDITY SIZE TANGIBILITY NDTAX MARKET_BOOK i.year i.SIC3) twostep