 |
 |
On 10/07/2010 04:34 AM, Kery Marc wrote: > > logit(psi.lim[i]) <- lpsi.lim[i] > lpsi.lim[i] <- min(999, max(-999, lpsi[i])) > lpsi[i] <- alpha.occ + beta.occ * something[i] > > BTW, I'd be happy to have comments on the "kosher-ness" of this. > > Regards -- Marc Truncating the linear predictor for the logit link at +/- 999 should really not affect the estimates since in that region of parameter space psi.lim is insensitive to changes in lspsi.lim, Some R code to illustrate: > inv.logit(-1000) == 0 [1] TRUE > inv.logit(-800) [1] 0 > inv.logit(-500) [1] 7.124576e-218 > inv.logit(-100) [1] 3.720076e-44 > inv.logit(-20) [1] 2.061154e-09 > inv.logit(-10) [1] 4.539787e-05 These are numbers from R, but the underlying issue is machine precision for real numbers, so the originators of that trick might have taken -999/999 as cutoffs because at that point the computer can't tell the difference. I would be cautious about the meaning of any logistic regression coefficients over 20 regardless of the model. Krzysztof ----------------------------------------------- Krzysztof Sakrejda-Leavitt Organismic and Evolutionary Biology University of Massachusetts, Amherst 319 Morrill Science Center South 611 N. Pleasant Street Amherst, MA 01003 work #: 413-325-6555 email: [log in to unmask] ----------------------------------------------- ------------------------------------------------------------------- This list is for discussion of modelling issues and the BUGS software. For help with crashes and error messages, first mail [log in to unmask] To mail the BUGS list, mail to [log in to unmask] Before mailing, please check the archive at www.jiscmail.ac.uk/lists/bugs.html Please do not mail attachments to the list. To leave the BUGS list, send LEAVE BUGS to [log in to unmask] If this fails, mail [log in to unmask], NOT the whole list