[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

From |
jverkuilen <jverkuilen@gc.cuny.edu> |

To |
<statalist@hsphsun2.harvard.edu> |

Subject |
RE: st: RE: Bootstrapping Conf Intervals - what do they mean? |

Date |
Wed, 4 Jun 2008 10:42:44 -0400 |

###Does that mean that the only "practical" advantage of the confidence interval is judging the precision of the study estimate?### Advantage compared to what? Leaving aside the issues with what p-values mean, the big advantage I see to interval estimates above hypothesis testing is the direct connection back to the original data. That is presumably what you care about, no? ###The above discussion suggests that it would be inaccurate to hint that the study's confidence interval brackets the population estimate.### Well it depends on what theory of probability you suscribe to. If you want to make probability statements about parameters as opposed to samples or functions of samples, i.e., estimators---and I would argue you probably do, but you may disagree---you are going to be going Bayesian. This is because in frequentist statistics the locus of randomness is on (usually hypothetical) repeated sampling from a population. Parameters are not random variables. *Estimators* are because they are functions of the data alone The quote from L. J. Savage: To make the Bayesian omelette, you have to break the Bayesian eggs. * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

- Prev by Date:
**st: R: growth rate** - Next by Date:
**st: growth rate** - Previous by thread:
**st: Stratify analysis - logistic regression with dummies** - Next by thread:
**st: How to define weights in gllamm** - Index(es):

© Copyright 1996–2015 StataCorp LP | Terms of use | Privacy | Contact us | What's new | Site index |