[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

From |
Dana Chandler <dchandler@gmail.com> |

To |
statalist@hsphsun2.harvard.edu |

Subject |
st: Usage of dprobit - std errors of marginal effects and use in low success samples |

Date |
Fri, 21 Aug 2009 11:12:17 -0500 |

Hello statalist, I have two questions about the usage of dprobit: the first regarding standard errors of marginal effects, the second on using marginal effects on phenomena that have extremely low successes. I recently employed a dprobit model to try and estimate the marginal effects. I noticed that when you divide the coefficients by the standard errors, you don't get the z-score at the right. In fact, that z-score is the same as if you had run a probit. How can one determine the significance level of these marginal effects (dFdx)s? When using the marginal effects, I know that the default is to estimate the marginal effects at the mean of all the covariates... since I have an extremely low success rate in the sample, all of these marginal effects are something on the order of x.x to the 10^(-9) or lower making interpretation difficult. If I have one or two variables that are really powerful in predicting success, might it be worthwhile to estimate the marginal effects at high values of that variable ? Would I get higher estimates of the other Xs if I condition on the X that is most important? Can statalisters suggest any other strategies? Also, although I don't think it's important, in all cases I've employed the "asis" option. Thanks in advance for all of your help, Dana Chandler * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

- Prev by Date:
**st: xtreg A B C , fe robust where R overall sq = 0** - Next by Date:
**Re: st: degenerate normal (mild annoyance)** - Previous by thread:
**st: xtreg A B C , fe robust where R overall sq = 0** - Next by thread:
**st: svyset for survey with oversampling** - Index(es):

© Copyright 1996–2017 StataCorp LLC | Terms of use | Privacy | Contact us | What's new | Site index |