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

From |
"Eva Poen" <eva.poen@gmail.com> |

To |
statalist@hsphsun2.harvard.edu |

Subject |
Re: st: nlcom question |

Date |
Fri, 11 Jul 2008 14:14:16 +0100 |

Hm, I might be missing something, but why do you do it in two steps? I.e. first alpha and then beta and gamma? Here is your example, using the auto data. Is this what you want? Hope this helps, Eva ***** sysuse auto regress price rep78 mpg foreign estimates store automodel local alp "_b[rep78]/(1+_b[rep78])" nlcom (alpha: `alp') (beta: _b[foreign]*(1-`alp')) (gamma: _b[mpg]*(1-`alp')), post scalar balpha=_b[alpha] scalar sealpha=_se[alpha] scalar bbeta=_b[beta] scalar sebeta=_se[beta] scalar bgamma=_b[gamma] scalar segamma=_se[gamma] estimates restore automodel outreg using myfile.out, coefastr se 10pct replace /* */ addstat("adj.Rsq", e(r2_a), "F test", e(F), /* */ "alpha", balpha, "se of alpha", sealpha, "beta", bbeta, "se of beta", sebeta, /* */ "gamma", bgamma, "se of gamma", segamma) adec(2) ****** 2008/7/11 Prof. Dr. Aysit Tansel <atansel@metu.edu.tr>: > Dear Statalist, > I am running the following regression after which I am trying to compute the > nonlinear combinations of the coefficients as defined in "alpha", "beta" and > "gamma". > in the "nlcom" command as given below. As you would notice, the"alpha" > computed in the first "nlcom" appears in the computations of "beta and > "gamma" in the subsequent "nlcom" commands. while this set of commands gives > me the correct values for "alpha" and the "standard error of alpha", there > is a problem with the subsequent computations. namely, the computed values > of "beta" and "gamma" are correct, but their standard errors are incorrect > since these computations treat > "alpha" as a scalar in these computations. > My first question is how to overcome this and the second question is how to > output the values of "alpha" "beta" and "gamma" and their standard errors > and/ or their t-ratios in the outreg file after the usual regression > statistics.... > Will be grateful for your help.... > > regress Y X1 X2 X3 > _estimates hold model7500, copy > nlcom alpha:_b[X1]/(1+_b[X1], post > scalar balpha=_b[alpha] > scalar sealpha=_se[alpha] > *scalar palpha=_p[alpha] > _estimates unhold model7500 > nlcom (beta: _b[X3]*(1-balpha)) (gamma: _b[X2]*(1-balpha)), post > scalar bbeta=_b[beta] > scalar sebeta=_se[beta] > scalar bgamma=_b[gamma] > scalar segamma=_se[gamma] > regress Y X1 X2 X3 > outreg using myfile.out, coefastr se 10pct replace /* > */ addstat("adj.Rsq", e(r2_a), "F test", e(F), /* > */ "alpha", balpha, "se of alpha", sealpha, "beta", bbeta, "se of beta", > sebeta, /* > */ "gamma", bgamma, "se of gamma", segamma) adec(2) > > > * * 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/

**References**:**st: nlcom question***From:*"Prof. Dr. Aysit Tansel" <atansel@metu.edu.tr>

- Prev by Date:
**st: nlcom question** - Next by Date:
**Re: st: suest with multinomial logit** - Previous by thread:
**st: nlcom question** - Index(es):

© Copyright 1996–2016 StataCorp LP | Terms of use | Privacy | Contact us | What's new | Site index |