Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

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

From |
Austin Nichols <austinnichols@gmail.com> |

To |
statalist@hsphsun2.harvard.edu |

Subject |
Re: st: deriving the variance-covariance matrix from point estimates and standard errors |

Date |
Thu, 23 Jun 2011 13:46:39 -0400 |

Michael <Michael.Monuteaux@childrens.harvard.edu>: Perhaps my example of "More detail may help" was not clear enough; for what purpose do you need the covariance of 2 estimates? If the data are publicly available, then you can download them and compute the point estimates and VCE yourself, no? It turns out that at least one of the special cases where I was thinking you did not need the whole VCE does not work with survey data--the case of a regression with two dummies and their interaction (for -regress-, you can recreate the whole VCE from the reported SEs but for -svy:reg- you cannot; it is the correction for clustering that breaks that particular trick, so the trick still works for weighted regression without a cluster-robust VCE). On Thu, Jun 23, 2011 at 12:01 PM, Monuteaux, Michael <Michael.Monuteaux@childrens.harvard.edu> wrote: > Thanks for responding. The data are from the NEISS-AIP survey, which is a nationally representative survey of injuries that were treated in U.S. emergency departments. The data are publically available through the CDC's website. For a given type of injury in a given demographic (eg, males), the website provides the estimated national number of injuries, the standard error of this estimate, and the number of survey cases that the estimate was based on. > > -----Original Message----- > From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Austin Nichols > Sent: Thursday, June 23, 2011 11:09 AM > To: statalist@hsphsun2.harvard.edu > Subject: Re: st: deriving the variance-covariance matrix from point estimates and standard errors > > Michael <Michael.Monuteaux@childrens.harvard.edu>: > This is only possible in a few special cases, without access to the > VCE of the original estimates, and the VCE is not typically reported > (if it were, you could just take a submatrix; if not you may have to > contact the original authors). More detail may help, e.g. are these > regression coefs for two dummies and you want to compute a test > statistic for the sum of their effects? > > On Thu, Jun 23, 2011 at 10:13 AM, Monuteaux, Michael > <Michael.Monuteaux@childrens.harvard.edu> wrote: >> Hello everyone, >> >> I have two point estimates and their corresponding standard errors from a survey, and I want to calculate the 2-by-2 variance-covariance matrix. Is there a way to do this in STATA? Please note, I do not have the original data, I just have the following info: >> The point estimate (in this case, it's the estimated number of cases who answered "yes" to a particular item in the survey) >> The standard error, and >> The number of records that the estimate was calculated from. >> >> Any help would be appreciated. Thanks!! >> >> Michael C. Monuteaux, Sc.D. >> Children's Hospital Boston * * 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/

**Follow-Ups**:**RE: st: deriving the variance-covariance matrix from point estimates and standard errors***From:*"Monuteaux, Michael" <Michael.Monuteaux@childrens.harvard.edu>

**References**:**st: deriving the variance-covariance matrix from point estimates and standard errors***From:*"Monuteaux, Michael" <Michael.Monuteaux@childrens.harvard.edu>

**Re: st: deriving the variance-covariance matrix from point estimates and standard errors***From:*Austin Nichols <austinnichols@gmail.com>

**RE: st: deriving the variance-covariance matrix from point estimates and standard errors***From:*"Monuteaux, Michael" <Michael.Monuteaux@childrens.harvard.edu>

- Prev by Date:
**Re: st: Complex Stata Code Examples** - Next by Date:
**RE: st: RE: looping up to a local macro** - Previous by thread:
**RE: st: deriving the variance-covariance matrix from point estimates and standard errors** - Next by thread:
**RE: st: deriving the variance-covariance matrix from point estimates and standard errors** - Index(es):