Hey Jens, It does on my computer, and detaching Zelig doesn't help. Look what happens: [1] 1033.52 Loading required package: MASS Loading required package: boot ## ## Zelig (Version 3.1-0, built: 2007-11-30) ## Please refer to http://gking.harvard.edu/zelig for full documentation ## or help.zelig() for help with commands and models supported by Zelig. ## Warning message: package 'MASS' was built under R version 2.6.2 [1] 1034 [1] 1034 [1] 1034 Unfortunately what you suggested doesn't help either. This is really weird. Any ideas? Jon 2008/3/18, Jens Hainmueller <jhainmueller at gmail.com>: -- Jon Bischof Graduate Student Department of Government Harvard University

Hi Kelly, No, updating my version of R did not help the problem. Also, loading the MASS and boot libraries separately, as Jens suggested, does not cause the same problem. Very curious. On an unrelated note, does anyone know if it's possible for the log-likelihood to go *down* after adding an explanatory variable? I tried to add a proxy for age of the program to my logit and the "value" out of optim on my unrestricted model was lower than the original! (Making a ratio test impossible, I think) Thoughts? Jon 2008/3/18, Bay, Kelly <Kelly_Bay at brown.edu>: -- Jon Bischof Graduate Student Department of Government Harvard University

The default value of "digits" option in R is 7 [1] 7 [1] 1234567 [1] 12345.67 When Zelig is loaded, it changes the value of "digits" option to 4 (see. Zelig/R/load.first.R ) Loading required package: MASS Loading required package: boot ## ## Zelig (Version 3.1-0, built: 2007-06-12) ## Please refer to http://gking.harvard.edu/zelig for full documentation ## or help.zelig() for help with commands and models supported by Zelig. ## [1] 4 [1] 12.34 [1] 1235 > ##Of course, you can change the "digits" option > options("digits" = 8) > 12345.678 [1] 12345.678 thanks, Ferdi Jens Hainmueller wrote: > John, > > This does look strange. Try to load MASS and boot separately and see if they > cause the same problem. If not, then it must be Zelig and I would contact > the maintainer of the package about this (Kosuke Imai). > > I just replicated your example on my machine and I do get the same issue: > > >> 12000.22993 >> > [1] 12000.23 > >> library(Zelig) >> > Loading required package: MASS > Loading required package: boot > ## > ## Zelig (Version 3.1-0, built: 2007-11-30) > ## Please refer to http://gking.harvard.edu/zelig for full documentation > ## or help.zelig() for help with commands and models supported by Zelig. > ## > >> 12000.22993 >> > [1] 12000 > >> options(digits=2) >> 12000.22993 >> > [1] 12000 > > Jens > > > >> -----Original Message----- >> From: gov2001-l-bounces at lists.fas.harvard.edu [mailto:gov2001-l- >> bounces at lists.fas.harvard.edu] On Behalf Of Jon Bischof >> Sent: Tuesday, March 18, 2008 9:10 AM >> To: gov2001-l at lists.fas.harvard.edu >> Subject: Re: [gov2001-l] zelig and decimals >> >> Hey Jens, >> >> It does on my computer, and detaching Zelig doesn't help. Look what >> happens: >> >> >>> 1033.52 #look it works! >>> >> [1] 1033.52 >> >>> library(Zelig) >>> >> Loading required package: MASS >> Loading required package: boot >> ## >> ## Zelig (Version 3.1-0, built: 2007-11-30) >> ## Please refer to http://gking.harvard.edu/zelig for full >> documentation >> ## or help.zelig() for help with commands and models supported by >> Zelig. >> ## >> Warning message: >> package 'MASS' was built under R version 2.6.2 >> >>> 1033.52 #what is going on? >>> >> [1] 1034 >> >>> detach(package:Zelig) >>> 1033.52 #still not working >>> >> [1] 1034 >> >>> options(scipen = 6) # controls scientific notation >>> options(digits = 2) # controls printed decimal places >>> 1033.52 #still not working >>> >> [1] 1034 >> >> Unfortunately what you suggested doesn't help either. This is really >> weird. Any ideas? >> >> Jon >> >> 2008/3/18, Jens Hainmueller <jhainmueller at gmail.com>: >> >>> detach(package:Zelig) >>> >>> does detach the package. Zelig doesn't do such rounding on my >>> >> machine: >> >>> > z.out1$coef >>> (Intercept) gdp capmob trade >>> 6.18129 -0.32360 1.42194 0.01985 >>> >>> Maybe check the digits settings: >>> >>> options(scipen = 6) # controls scientific notation >>> options(digits = 1) # controls printed decimal places >>> >>> hope this helps. >>> >>> jens >>> >>> >>> >>> >>> >>> > -----Original Message----- >>> > From: gov2001-l-bounces at lists.fas.harvard.edu [mailto:gov2001-l- >>> > bounces at lists.fas.harvard.edu] On Behalf Of Jon Bischof >>> > Sent: Tuesday, March 18, 2008 7:29 AM >>> > To: gov2001-l at lists.fas.harvard.edu >>> > Subject: [gov2001-l] zelig and decimals >>> >>> > Hi all, >>> >>> > It appears that Zelig forces R to round numbers fairly strictly >>> >> such >> >>> > that 466.543 will always be 466, even if you try to get R to keep >>> >> more >> >>> > decimals. Why does Zelig do this? Is there a way to turn this off >>> >> in >> >>> > Zelig or to detach the library? >>> >>> > -- >>> > Jon Bischof >>> > Graduate Student >>> > Department of Government >>> > Harvard University >>> >>> >>>> _______________________________________________ >>>> >>> > gov2001-l mailing list >>> > gov2001-l at lists.fas.harvard.edu >>> > http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l >>> >>> _______________________________________________ >>> gov2001-l mailing list >>> gov2001-l at lists.fas.harvard.edu >>> http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l >>> >>> >>> >> -- >> Jon Bischof >> Graduate Student >> Department of Government >> Harvard University >> _______________________________________________ >> gov2001-l mailing list >> gov2001-l at lists.fas.harvard.edu >> http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l >> > > _______________________________________________ > gov2001-l mailing list > gov2001-l at lists.fas.harvard.edu > http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l > -- Ferdinand Alimadhi Manager of Statistical Programming Harvard University The Institute for Quantitative Social Science (617) 496-0187 falimadhi at iq.harvard.edu

I had this issue before, too. it appears that digits option has to do with significant figures, not with places after the decimal point. So 12000.1232 to two sig-figs is indeed 12000. sprintf("%.2f", value) can be used in write-ups to output exact number of digits after the decimal point. -alexei On 03/18/2008 11:39 AM, Jon Bischof wrote: