Dear all, Is there any function for calculating confidence limits for coefficients in an lm() object? I know of the confint() function in the MASS library working very well on my binomial GLMs and I have tried it (using glm () , family=gaussian) but it gives NAs according to below. Does the confint() function not accept gaussian GLMs? Could there be convergence problems in the GLM? Note the very low R2-value. Could the Hauck & Donner phenomenon discussed in V & R (1999) occur in a gaussian GLM? I guess not and I have tried with different epsilon but it does not change anything. I use R 1.1.0 on Windows 98. Sorry to bother you about this but earlier some R users had problem using the Up arrow key to get the earlier written text rows. I deleted those mails because I usually use Windows NT. Now however I?m away from my office using R on Windows 98 (something I didn\'t plan to do). Could someone who have had this problem please tell me how to do to be able to use the Up arrow key again. Thanks for all hints! Sincerely, Tord Sn?ll> glm.spe.var<- glm(OSPEABUN~ V1+V2+V3+V4+V5+V6+V7+V8,family=gaussian, data=R)> lm.spe.var<- lm(OSPEABUN~V1+V2+V3+V4+V5+V6+V7+V8,data=R)> summary(lm.spe.var)Residuals: Min 1Q Median 3Q Max -0.854243 -0.353655 -0.212623 -0.004446 2.729957 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 0.274985 0.028895 9.517 < 2e-16 *** V1 0.145041 0.041001 3.537 0.000445 *** V2 -0.002472 0.047220 -0.052 0.958276 V3 -0.016563 0.046899 -0.353 0.724120 V4 -0.065702 0.029530 -2.225 0.026573 * V5 0.023614 0.031004 0.762 0.446665 V6 0.130698 0.031565 4.141 4.12e-05 *** V7 -0.009698 0.042213 -0.230 0.818398 V8 -0.041318 0.031670 -1.305 0.192672 --- Signif. codes: 0 `***\\\' 0.001 `**\\\' 0.01 `*\\\' 0.05 `.\\\' 0.1 ` \\\' 1 Residual standard error: 0.6257 on 460 degrees of freedom Multiple R-Squared: 0.07881, Adjusted R-squared: 0.06279 F-statistic: 4.92 on 8 and 460 degrees of freedom, p-value: 7.505e-006> summary(glm.spe.var)Deviance Residuals: Min 1Q Median 3Q Max -0.854243 -0.353655 -0.212623 -0.004446 2.729957 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 0.274985 0.028895 9.517 < 2e-16 *** V1 0.145041 0.041001 3.537 0.000445 *** V2 -0.002472 0.047220 -0.052 0.958276 V3 -0.016563 0.046899 -0.353 0.724120 V4 -0.065702 0.029530 -2.225 0.026573 * V5 0.023614 0.031004 0.762 0.446665 V6 0.130698 0.031565 4.141 4.12e-05 *** V7 -0.009698 0.042213 -0.230 0.818398 V8 -0.041318 0.031670 -1.305 0.192672 --- Signif. codes: 0 `***\\\' 0.001 `**\\\' 0.01 `*\\\' 0.05 `.\\\' 0.1 ` \\\' 1 (Dispersion parameter for gaussian family taken to be 0.3915403) Null deviance: 195.52 on 468 degrees of freedom Residual deviance: 180.11 on 460 degrees of freedom AIC: 902.11 Number of Fisher Scoring iterations: 2> confint(glm.spe.var, level=0.95)Waiting for profiling to be done... 2.5 % 97.5 % (Intercept) NA NA V1 NA NA V2 NA NA V3 NA NA V4 NA NA V5 NA NA V6 NA NA V7 NA NA V8 NA NA -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
On Mon, 14 Aug 2000, Snall, Tord wrote:> > Dear all, > > Is there any function for calculating confidence limits > for coefficients in an lm() object? I know of the > confint() function in the MASS library working very > well on my binomial GLMs and I have tried it (using glm > () , family=gaussian) but it gives NAs according to > below. Does the confint() function not accept gaussian > GLMs? Could there be convergence problems in the GLM? > Note the very low R2-value. Could the Hauck & Donner > phenomenon discussed in V & R (1999) occur in a > gaussian GLM? I guess not and I have tried with > different epsilon but it does not change anything.The distribution of the coefficients for the Gaussian case is exact t, so you don't need any of this, just the textbook formulae. Given that, I suspect no one ever checked whether it works.> I use R 1.1.0 on Windows 98. > Sorry to bother you about this but earlier some R users > had problem using the Up arrow key to get the earlier > written text rows. I deleted those mails because I > usually use Windows NT. Now however I’m away from my > office using R on Windows 98 (something I didn\'t plan > to do). Could someone who have had this problem please > tell me how to do to be able to use the Up arrow key > again.Delete .Rhistory. -- Brian D. Ripley, ripley at stats.ox.ac.uk Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ University of Oxford, Tel: +44 1865 272861 (self) 1 South Parks Road, +44 1865 272860 (secr) Oxford OX1 3TG, UK Fax: +44 1865 272595 -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
> > Dear all, > > Is there any function for calculating confidence limits > for coefficients in an lm() object? I know of the > confint() function in the MASS library working very > well on my binomial GLMs and I have tried it (using glm > () , family=gaussian) but it gives NAs according to > below. Does the confint() function not accept gaussian > GLMs? Could there be convergence problems in the GLM? > Note the very low R2-value. Could the Hauck & Donner > phenomenon discussed in V & R (1999) occur in a > gaussian GLM? I guess not and I have tried with > different epsilon but it does not change anything.The software was never tested for this case as it seemed a little pointless since the exact calculations can be done so simply. (you might like to do it as a homework exercise...) I'll look at putting in a switch to handle the case for future releases. I suspect it will not work too well, though, as the method uses spline interpolation to the signed square-root of the profile function, which in this case is exactly linear. I think cubic spline interpolation of straight lines causes some numerical problems. This is just a quick response and you might like to check it.> > I use R 1.1.0 on Windows 98. > Sorry to bother you about this but earlier some R users > had problem using the Up arrow key to get the earlier > written text rows. I deleted those mails because I > usually use Windows NT. Now however I?m away from my > office using R on Windows 98 (something I didn\'t plan > to do). Could someone who have had this problem please > tell me how to do to be able to use the Up arrow key > again. > > Thanks for all hints! > > > Sincerely, > Tord Sn?ll > > > > > glm.spe.var<- glm(OSPEABUN~ V1+V2+V3+V4+V5+V6+V7+V8, > family=gaussian, data=R) > > lm.spe.var<- lm(OSPEABUN~V1+V2+V3+V4+V5+V6+V7+V8, > data=R) > > summary(lm.spe.var) > > > Residuals: > Min 1Q Median 3Q Max > -0.854243 -0.353655 -0.212623 -0.004446 2.729957 > > Coefficients: > Estimate Std. Error t value Pr(>|t|) > (Intercept) 0.274985 0.028895 9.517 < 2e-16 *** > V1 0.145041 0.041001 3.537 0.000445 *** > V2 -0.002472 0.047220 -0.052 0.958276 > V3 -0.016563 0.046899 -0.353 0.724120 > V4 -0.065702 0.029530 -2.225 0.026573 * > V5 0.023614 0.031004 0.762 0.446665 > V6 0.130698 0.031565 4.141 4.12e-05 *** > V7 -0.009698 0.042213 -0.230 0.818398 > V8 -0.041318 0.031670 -1.305 0.192672 > --- > Signif. codes: 0 `***\\\' 0.001 `**\\\' 0.01 `*\\\' > 0.05 > `.\\\' 0.1 ` \\\' 1 > > Residual standard error: 0.6257 on 460 degrees of > freedom > Multiple R-Squared: 0.07881, Adjusted R-squared: > 0.06279 > F-statistic: 4.92 on 8 and 460 degrees of freedom, > p-value: 7.505e-006 > > > summary(glm.spe.var) > > > Deviance Residuals: > Min 1Q Median 3Q Max > -0.854243 -0.353655 -0.212623 -0.004446 2.729957 > > Coefficients: > Estimate Std. Error t value Pr(>|t|) > (Intercept) 0.274985 0.028895 9.517 < 2e-16 *** > V1 0.145041 0.041001 3.537 0.000445 *** > V2 -0.002472 0.047220 -0.052 0.958276 > V3 -0.016563 0.046899 -0.353 0.724120 > V4 -0.065702 0.029530 -2.225 0.026573 * > V5 0.023614 0.031004 0.762 0.446665 > V6 0.130698 0.031565 4.141 4.12e-05 *** > V7 -0.009698 0.042213 -0.230 0.818398 > V8 -0.041318 0.031670 -1.305 0.192672 > --- > Signif. codes: 0 `***\\\' 0.001 `**\\\' 0.01 `*\\\' > 0.05 > `.\\\' 0.1 ` \\\' 1 > > (Dispersion parameter for gaussian family taken to be > 0.3915403) > > Null deviance: 195.52 on 468 degrees of freedom > Residual deviance: 180.11 on 460 degrees of freedom > AIC: 902.11 > > Number of Fisher Scoring iterations: 2 > > > confint(glm.spe.var, level=0.95) > Waiting for profiling to be done... > 2.5 % 97.5 % > (Intercept) NA NA > V1 NA NA > V2 NA NA > V3 NA NA > V4 NA NA > V5 NA NA > V6 NA NA > V7 NA NA > V8 NA NA > > > > -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- > r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html > Send "info", "help", or "[un]subscribe" > (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch > _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._ >-- Bill Venables, Statistician, CMIS Environmetrics Project CSIRO Marine Labs, PO Box 120, Cleveland, Qld, AUSTRALIA. 4163 Tel: +61 7 3826 7251 Email: Bill.Venables at cmis.csiro.au Fax: +61 7 3826 7304 http://www.cmis.csiro.au/bill.venables/ -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._