0
$\begingroup$

Your task is to investigate how fertility affects labour supply. In particular, you will examine how a woman's labour supply falls when she has more than two children. Your answers should include your regression results. Remember to conduct the relevant tests to support your results as well. Furthermore, use robust standard errors in all your regressions.

What does it mean that I should use robust standard errors in all my regressions?

$\endgroup$
  • 1
    $\begingroup$ If this is from a class, have you not covered this topic in your reading or lectures? $\endgroup$ – AdamO Feb 4 at 20:28
  • $\begingroup$ I answered a question recently in which I explained "Robust" standard error (also called the "sandwich" standard errors or "Huber-white" standard errors): stats.stackexchange.com/questions/389808/… $\endgroup$ – StatsStudent Feb 4 at 22:44
1
$\begingroup$

A robust standard error is a different way of calculating the standard error of a regression coefficient in a regression model. It is also referred to as a sandwich standard error, an Eicker-White standard error, a Huber standard error, a heteroscedasticity consistent standard error, and probably a few other names. Unlike the default model-based way of calculating standard errors, the robust standard error uses the actual squared residuals instead of expected variance. Most software have implementations of it, so make good use of your search tool and documentation.

$\endgroup$
  • 1
    $\begingroup$ Would be improved by an explanation of why the sammich estimator should be preferred to the 'default model-based' estimator. $\endgroup$ – Alexis Feb 4 at 21:16
  • 1
    $\begingroup$ @Alexis but that wasn't the question, and that question (and a host of good answers) are already all over the site. $\endgroup$ – AdamO Feb 4 at 21:17
  • 1
    $\begingroup$ I do not see that you addressed the should use portion of the question. (Though you did address the what portion. :) $\endgroup$ – Alexis Feb 4 at 21:19
  • 1
    $\begingroup$ @Alexis I read "What does it mean that I should use robust standard errors in all my regressions?" as "What is robust standard error?". Not "Why should it be preferred to model-based?". $\endgroup$ – AdamO Feb 4 at 21:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.