Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | Nick Cox <njcoxstata@gmail.com> |
To | statalist@hsphsun2.harvard.edu |
Subject | Re: st: Re: Transforming Inflation |
Date | Fri, 25 Mar 2011 19:32:12 +0000 |
Well, have you tried plotting it? twoway function (x/100)/((100 +x)/100), ra(-100 100) is monotonic, but not at all symmetric about 0. If this were my problem, I would prefer my previous suggestion. On Fri, Mar 25, 2011 at 7:16 PM, ajjee <ajjee1@yahoo.com> wrote: > Thanks Nick > > You are absolutely right about the second formula....i was computing the > wrong. The correct formula is > transformed_inflation=(inflation/100)/((100 +inflation)/100), as you > mentioned. And this has no sign problems. > > Thanks again > > ajjee > > > > On 25 March 2011 19:57, Nick Cox <njcoxstata@gmail.com> wrote: > Deflating inflation! Can you do it for real too? > > Applying that transform is indeed quite absurd for negative values; > note that it is undefined for x = -1. I don't think there is any > singularity at 1% deflation. > > A generalisation that makes the transform symmetric about zero is > > sign(x) * abs(x)/(1 +abs(x)) > > By intent, this is monotonic and preserves sign, which seem > economically reasonable too. > > I discuss a bundle of related issues in > > Cox, N.J. 2011. Stata tip 96: Cube roots. > The Stata Journal 11(1): 149-154. > > I make a case for cube roots being the simplest shape-changing > transformation that preserves sign and is applicable to negative, zero > and positive values alike. > > Shouldn't the second formula be > > transformed_inflation=(inflation/100)/((100 +inflation)/100) > > On Fri, Mar 25, 2011 at 6:40 PM, ajjee <ajjee1@yahoo.com> wrote: > >> My question is not related to Stata but I have a technical problem in >> computing a variable. In estimation, normally we transform our inflation >> variable to reduce the influence of extreme observations by the formula: >> >> transformed_inflation=(inflation)/(1+inflation) or sometime >> transformed_inflation=(inflation/100)/((1+inflation)/100) >> >> But if the value of the variable is negative, say (-1.0666355), the >> resultant value is (16.00701) by the first formula which is misleading. >> What >> should be done in this situation? > > > > > -- > View this message in context: http://statalist.1588530.n2.nabble.com/Transforming-Inflation-tp6208771p6208877.html > Sent from the Statalist mailing list archive at Nabble.com. > * > * 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/ > * * 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/